University of Southern California Dissertations and Theses

Intricate microfluidic devices for biopharmaceutical processes: forging ahead with additive manufacturing

(USC Thesis Other)

Intricate microfluidic devices for biopharmaceutical processes: forging ahead with additive manufacturing

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content INTRICATE MICROFLUIDIC DEVICES FOR BIOPHARMACEUTICAL PROCESSES:
FORGING AHEAD WITH ADDITIVE MANUFACTURING
by
Wan-Zhen Sophie Lin
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(CHEMICAL ENGINEERING)
DECEMBER 2022
Copyright 2022 Wan-Zhen Sophie Lin
Dedication
I dedicate my dissertation to my parents for their love and encouragement. Being 6,773 miles away from
home, I could not have gotten this far without your help at every turn. In the face of adversity, you
reminded me to focus on personal growth and move forward with the lessons learned. Thank you for
giving me the strength to continuously learn, change, and challenge myself.
ii
Acknowledgements
First and foremost, I would like to thank my advisor, Prof. Noah Malmstadt. In 2016, my undergradu-
ate sophomore year, Prof. Malmstadt’s early acceptance of me marked the start of this fruitful journey,
for which I am eternally grateful. His encouragements were always accompanied with positive sugges-
tions, guiding me through thick and thin. Most importantly, I am deeply grateful to him for facilitating
such a positive learning environment, where I was trusted with the freedom to explore different research
approaches at my own pace. The opportunity to try out new ideas and overcome the hurdles proved
invaluable to my development as a scientist. Thank you for your endless patience, unfaltering positive
attitude, and tremendous support in my scientific endeavor.
Next, I would like to acknowledge Prof. Richard Roberts, Prof. Cristina Zavaleta, Prof. Malancha
Gupta, Prof. Megan McCain, and Prof. Nick Graham for being on my dissertation and qualifying com-
mittee. Thank you for your time, insightful remarks, and perceptive questions, which inspired me to dig
deeper. In particular, I would like to extend a gracious thank you to Prof. Roberts for his guidance in
our collaborative research projects and permission to carry out experiments in his laboratory. He is an
invaluable source of biophysical chemistry knowledge, and without his generous hospitality, my research
projects would not have progressed so smoothly.
I would also like to thank all the present and past members of the Malmstadt Lab, my research collab-
orators, and colleagues who helped me and supported me throughout my PhD. My academic journey has
not been a simple journey of progress, and I am extremely fortunate to have you with me through the ups
iii
and downs. Special appreciations to Dr. William Evenson and Justin Ong for their close collaborations and
expertise; W. Kristian Vu Bostic, Kenmond Pang, and Alex Czaja, whom I’ve had the privilege to mentor
during my PhD, for their assistance and enthusiasm; Dr. Sepehr Maktabi, Dr. Kaori Noridomi, and Dr. Lu
Wang for their constructive advice. I will treasure these memories wherever my journey takes me.
Last but not least, I would like to express my profound gratitude and appreciation to my family. My
greatest thanks go to my parents, Yung-Yang Lin and Shu-Fang Ho, and my sister, Chia-An Lin, for their
love and immense support in every conceivable manner. I thank Brian Feng for walking alongside me
throughout my PhD. Thank you for catching me when I fall, holding me up when I feel weak, and loving
me through it all. Your company has made my academic journey truly enjoyable.
iv
TableofContents
Dedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii
Chapter 1: Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Microfluidics for Biopharmaceutical Applications . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 Microfluidic Synthesis in Nanomedicine . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.2 Microfluidics for Affinity Reagent Development . . . . . . . . . . . . . . . . . . . . 4
1.2 Towards 3D Printed Microfluidics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.2.1 The Rise of 3D Printing for Microfluidics . . . . . . . . . . . . . . . . . . . . . . . . 5
1.2.2 3D Printing Technologies for Microfluidics . . . . . . . . . . . . . . . . . . . . . . . 7
Chapter 2: Microfluidic Liposome Production and Concurrent Loading of Drug Simulants . . . . . 11
2.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.3 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.3.1 Device Fabrication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.3.2 Lipid Mixture and Hydration Buffer Preparation . . . . . . . . . . . . . . . . . . . . 15
2.3.3 Liposome Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.3.4 Hydrophilic Drug Loading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.3.5 FITC-dextran Diffusivity Calculation . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.3.6 Hydration Buffer Recycling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.3.7 Hydrophobic Drug Loading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.3.8 Hydrophilic and Hydrophobic Drug Loading . . . . . . . . . . . . . . . . . . . . . . 18
2.3.9 Statistical Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.4 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Chapter 3: 3D-Printed Microfluidic Device for High-Throughput Production of Lipid Nanoparticles
Incorporating SARS-CoV-2 Spike Protein mRNA . . . . . . . . . . . . . . . . . . . . . . 28
3.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.2 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.3 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
v
3.3.1 Preparation of Lipid Mixtures and mRNA Encoding SARS-CoV-2 Spike Protein . . 33
3.3.2 Device Design and Fabrication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.3.3 Production of Lipid Nanoparticles . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.3.4 LNP Characterization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.3.5 COMSOL Multiphysics Flow Simulation . . . . . . . . . . . . . . . . . . . . . . . . 40
3.4 Results & Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Chapter 4: Kinetic Off-Rate Selections Using a 3D-Printed Microfluidic Device . . . . . . . . . . . . 52
4.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.2 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.3 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.3.1 MFED Design and Fabrication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.3.2 Ligand Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.3.3 Target Protein . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.3.4 Radioactive Binding Assays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.3.5 Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
4.3.6 PCR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
4.3.7 Band Intensity Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.4 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
Chapter 5: Compatibility of Popular Three-Dimensional Printed Microfluidics Materials with
In-Vitro Enzymatic Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
5.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
5.2 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
5.3 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
5.3.1 3D Printing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
5.3.2 PCR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
5.3.3 Transcription . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
5.3.4 Translation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
5.3.5 Reverse Transcription . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
5.4 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
5.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
Chapter 6: Conclusions and Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
vi
ListofFigures
1 The microfluidic hydrodynamic focusing device. (A) Hydration buffer solution and lipid
solution in IPA entered the device through fluidic access points a and b, respectively. The
liposome product was collected from fluidic access point c. (B) The height of the channel
was 50µm, and the width was 60 or 80µm. . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2 FITC-dextran loading. (A) 4kDa, 20kDa, and 40kDa FITC-dextran loading efficiencies
corresponding to different FRRs. The lipid mixture inlet flow rate was controlled at
3µL/min. (B) Mean liposome radii corresponding to different IPA concentrations in the
hydration buffer inlet. (C) FITC-dextran loading efficiency at varying TFR and a constant
FRR of 30. (D) Mean liposome radii at varying TFR and a constant FRR of 30. . . . . . . . . 20
3 FITC-dextran diffusivity and loading efficiency. (A) 4kDa, 20kDa, and 40kDa FITC-dextran
diffusivities were approximated. (B) FITC-dextran loading efficiency corresponding to
diffusivity. The error bars presented the uncertainties of the data obtained by error
propagation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4 Reusing product filtrate as the hydration buffer for liposome fabrication. Three trials were
done at FRR of 20, 30, and 50. The lipid mixture inlet flow rate was controlled at 3 µL/min
while the FRR varied. (A) FITC-dextran loading efficiency corresponding to the number of
times the collected filtrate was reused as a hydration buffer inlet. (B) Mean liposome radii
corresponding to the number of times the filtrate was reused as a hydration buffer inlet. . 22
5 FITC-dextran loading at varying %IPA and FITC-dextran concentration in the hydration
buffer. The lipid mixture inlet flow rate was controlled at 3 µL/min. (A) FITC-dextran
loading efficiencies corresponding to different IPA concentrations in the hydration buffer
inlet. (B) Mean liposome radii corresponding to different IPA concentrations in the
hydration buffer inlet. (C) FITC-dextran loading efficiencies corresponding to different
FITC-dextran concentrations in the hydration buffer inlet. (D) Mean liposome radii
corresponding to different FITC-dextran concentrations in the hydration buffer inlet. . . . 23
vii
6 Nile red loading. (A) Nile red loading efficiency versus FRR. The lipid mixture inlet flow
rate was controlled at 3µL/min. (B) Mean liposome radius versus FRR when Nile red was
loaded. The lipid mixture inlet flow rate was controlled at 3 µL/min. (C) Nile red loading
efficiencies versus TFR at a constant FRR of 30. (D) Mean liposome radius versus TFR at
a constant FRR of 30 when Nile red was loaded. The error bar for each Nile red loading
efficiency showed the standard deviation obtained by at least three repeated trials of the
experiment. The error bars for liposome radii showed the polydispersity measured by the
Dynamic Light Scattering. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
7 Concurrent loading of 4kDa FITC-dextran and Nile red on liposomes. FRR was varied
from 10 to 50 while maintaining a constant lipid mixture inlet flow rate of 3 µL/min. (A)
FITC-dextran loading efficiency when loaded into liposomes alone versus concurrent
loading of both drug simulants. (B) Nile red loading efficiency when loaded into liposomes
alone versus concurrent loading of both drug simulants. (C) Mean liposome radii under
different loading conditions. The error bar for each loading efficiency showed the standard
deviation obtained by at least three repeated trials of the experiment. The error bars for
liposome radii showed the polydispersity measured by the Dynamic Light Scattering. . . . 26
8 Plasmid insert sequence encoding SARS-CoV-2 spike protein and PCR primer sequences.
Yellow: T7 promoter, blue: 3’ untranslated region, hunter green: start codon, brown:
signal peptide, orange: spike protein, red: stop codon, magenta: 3’ untranslated region,
light green: BbsI restriction site. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
9 Gel electrophoresis results for (A) PCR and (B) transcription products from pDNA
encoding SARS-CoV-2 spike protein. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
10 (A) Isometric view,(B) front view, and(C) side view of the OSEM device featuring a 4-way
sheath flow channel followed by a downstream SHM for LNP production. . . . . . . . . . . 36
11 The OSEM design in drawings showing detailed design parameters. Unit: mm. . . . . . . . 37
12 Photograph of a 4-way sheath flow demonstrated using water with (inlet flow rate = 0.25
mL/min) and without (inlet flow rate = 5 mL/min) red food dye (FRR = 20). . . . . . . . . . 37
13 Photographs of 3D-printed microfluidic devices featuring (A) a 2-way sheath flow channel
followed by a downstream SHM and(B) a 4-way sheath flow channel without downstream
mixers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
14 (A) Size and(B) mRNA encapsulation efficiency of LNPs produced using a 4-way sheath
flow channel with downstream SHM device (4-Way Sheath), a 2-way sheath flow channel
with downstream SHM device (2-Way Sheath), and a 4-way sheath flow channel without
the downstream SHM device (No Mixer). LNPs were produced at a total inlet mRNA flow
rate of 4 mL/min, mRNA inlet concentration of 5 ng/µL, and FRR of 20. Error bars show
standard deviations across> 3 repeated trials. Photographs of the 2-Way Sheath Flow
device and the No Mixer device are shown in Figure 13. . . . . . . . . . . . . . . . . . . . . 41
viii
15 COMSOL Multiphysics simulation results showing ethanol concentration in water and
outlet flow velocity within hydrodynamic focusing channels. The FRR was set to 20 and
the total water inlet flow rate was set to 5 mL/min. (A) & (E) Images of the 2-way and
4-way sheath flow region of the device, respectively. Ethanol concentration results for
a 2-way sheath flow channel in (B) the x-y plane at the center of the channel, (C) an
orthogonal view of showing concentration profiles in x-z planes, and(D) the x-z plane at
outlety =− 3 mm with flow velocity contours. Ethanol concentration results for a 4-way
sheath flow channel in (F) the x-y plane at the center of the channel,(G) orthogonal view
showing concentration profiles in x-z planes, and(H) the x-z plane at outlety =− 3 mm
with flow velocity contours. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
16 COMSOL Multiphysics simulation results showing ethanol/water hydrodynamic focusing
at 5 µ L/min total water inlet flow rate. The FRR was set to 20. (A) & (D) Results of a
2-way and a 4-way sheath flow regions of the device respectively, in the x-y plane. (B) &
(E) Results for a 2-way and a 4-way sheath flow channel, respectively, in orthogonal view
showing slices of the concentration profile in x-z planes. (C) & (E) Results for a 2-way
and a 4-way sheath flow channel, respectively, in the x-z plane at outlety =− 3 mm with
flow velocity contours. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
17 DLS size (column 1), zeta potential (column 2), and encapsulation efficiency (column 3)
results for LNPs produced under various conditions. (A) Results for LNPs produced at
mRNA inlet concentration ranging from 5 to 40 ng/µL, at constant FRR of 20 and inlet
mRNA flow rate of 4 mL/min. (B) Results for LNPs produced at inlet mRNA flow rate
ranging between 2− 6 mL/min (small-scale) and 24− 60 mL/min (high throughput),
at constant FRR of 20 and inlet mRNA concentration of 10 ng/µL. LNPs produced at 5
mL/min mRNA inlet flow rate (highlighted) were imaged using cryo-TEM as shown in
Figure 18. (C) Results for LNPs produced at FRR ranging from 5 to 30, at constant total
flow rate of 4.2 mL/min, inlet mRNA concentration of 10 ng/ µL. Error bars of standard
deviations among> 3 repeated trials were plotted. . . . . . . . . . . . . . . . . . . . . . . . 48
18 Cryo-TEM images of LNPs produced by the 3D-printed microfluidic device at FRR = 20,
inlet mRNA concentration = 10 ng/µL, and inlet mRNA flow rate = 5 mL/min. (A) LNPs in
20mM Tris-Cl buffer (pH 7.5). (B) LNPs stained with 1% uranyl acetate (pH 4.0). . . . . . . 49
19 DLS size measurment results of LNPs on the day they were produced vs 30 days after
storage at 4°C. The LNPs were produced at FRR = 20, inlet mRNA flow rate of 5 mL/min,
and inlet mRNA concentration of 10 ng/µL. Error bars show standard deviations among>
3 repeated trials. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
20 Design of the MFED. (A) Schematic illustration of bead washing on the MFED. (B) Picture
of the assembled MFED. (C) Picture of the disassembled MFED. 1 cm scale bar indicated. . 55
21 Characterizing and modeling E1 peptide-mRNA fusion binding at various flow rates.
The percent bound increases as the flow rate decreases and plateaus at 73%. Error bars
represent the standard deviation of percent bound over three trials. The best-fit model
givesk
on
= (2.5± 4)× 10
4
M
− 1
s
− 1
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
ix
22 (A) MFED loading and washing E1 (red circle) and Pep2 (green circle) mRNA-peptide
fusions. Error bars indicate the standard deviation of three trials. Initial binding for E1
(52%) shows little decay during washing, whereas Pep2 binding (12± 1%) washes out with
a first order rate constant k
off
= 9± 2× 10
− 3
s
− 1
. The ratio of E1:Pep2 bound (blue
square) plateaus after 1,000 seconds when the Pep2 binding reaches background binding
(1.4± 0.3%). (B) Radiolabeled manual competitor-based off-rate measurement of Pep2
(green triangle). Beads containing Bcl-x
L
were exposed to radiolabeled Pep2, washed, and
resuspended with 100X free Bcl-x
L
. Timepoints were taken after the resuspension in free
Bcl-x
L
and the percent of Pep2 retained on the beads was recorded and normalized such
that the initial binding percentage was 100%. The curve was fit as a single exponential
decay using a nonlinear fit in GraphPad Prism. The k
off
was 1.1× 10
− 2
s
− 1
, and a
minimum percent bound of 19%. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
23 Dilution and PCR of 1:1 mixture of E1 and Pep2 PCR products. The gel image indicates
that after 7 cycles, the band intensities for E1 and Pep2 are comparable, and this continues
to be true through 22 cycles. This implies that there is minimal PCR bias. . . . . . . . . . . 65
24 Manual versus MFED enrichment of E1 over Pep2. (A) Defined standard mixtures of ligated
E1 and Pep2 were prepared and reverse transcribed. The sample was then amplified by
PCR and run on agarose gels. (B) The band intensity ratios of the PCR amplified standard
mixtures were calculated using Image Studio Lite to build a standard curve. (C) Selection
was performed on various starting ratios of E1:Pep2 (1:2, 1:38, 1:67, 1:135, and 1:292)
and the resulting PCR products were run on agarose gels. (D) The band intensities were
measured and converted into molar ratios using the standard curve. MFED selection
consistently outperformed manual selection, averaging 13-fold enrichment compared to
5-fold enrichment observed with manual selection (* indicates p < 0.05). . . . . . . . . . . . 66
25 Illustration of a 3D-printed cone placed in a microcentrifuge tube. . . . . . . . . . . . . . . 71
26 (A) PCR yield (normalized to no BSA, no parts sample) at various BSA wt%. 0.4 wt%
BSA (highlighted in grey) was chosen for the following experiments. (B) PCR yield
(normalized to no BSA, no parts sample) when incubated with different 3D printed
materials, performed with and without BSA (0.4 wt%). BSA significantly improved PCR
yields when 3D printed parts were present (p< 0.05). Gel images shown in Figure 27. . . . 74
27 PCR Results. (A) A gel image showing PCR outcomes at different wt% BSA concentrations.
(B) A gel image showing PCR outcomes when incubated with 3D-printed parts with or
without BSA. Ladder: 100bp DNA Ladder, New England BioLabs. Electrophoresis Buffer:
1X SB. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
28 (A) Transcription yield (normalized to no BSA, no parts sample) at various BSA wt%. 0.4
wt% BSA (highlighted in grey) was chosen for the following experiments. (B) Transcription
yield (normalized to no BSA, no parts sample) when incubated with different 3D printed
materials, performed with and without BSA (0.4 wt%). BSA significantly improved the
transcription yield when 3D printed parts were present (p < 0.05). The yields of printed
parts without BSA and PEGDA-based resin with 0.4 wt% BSA were below the detection
limit (Figure 29). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
x
29 Transcription Results. (A) A gel image showing transcription outcomes at different wt%
BSA concentrations. (B) A gel image showing transcription outcomes when incubated
with 3D-printed parts with or without BSA. (C) A photo of the samples collected in
Eppendorf tubes showing gelation at high BSA concentrations. Electrophoresis Buffer: 1X
TBE. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
30 (A) Translation yield (normalized to no BSA, no parts sample) at various BSA wt%. 0.6 wt%
BSA (highlighted in grey) was chosen for the subsequent experiments. (B) Translation
yield (normalized to no BSA, no parts sample) when incubated with different 3D printed
materials, performed with and without BSA (0.6 wt%). BSA did not significantly improve
translation yield when 3D printed parts were present. . . . . . . . . . . . . . . . . . . . . . 77
31 (A) Reverse transcription yield (normalized to no BSA, no parts sample) at various BSA
wt%. 0.4 wt% BSA (highlighted in grey) was chosen for the subsequent experiments.
(B) Transcription yield (normalized to no BSA, no parts sample) when incubated
with different 3D printed materials, performed with and without BSA (0.6 wt%). BSA
significantly improved the reverse transcription yield when 3D printed parts were present
(p< 0.05). The yields of PEGDA-based resin without BSA and Pro3dure GR-1 without
BSA were below the detection limit (Figure 32). . . . . . . . . . . . . . . . . . . . . . . . . 78
32 Reverse Transcription Results. (A) A gel image showing reverse transcription outcomes
at different wt% BSA concentrations. (B) A gel image showing reverse transcription
outcomes when incubated with 3D-printed parts with or without BSA. Ladder: 100bp
DNA Ladder, New England BioLabs. Electrophoresis Buffer: 1X SB. . . . . . . . . . . . . . 79
33 Transcription yield (normalized to no BSA, no parts sample) when incubated with parts vs
with leachate isolated from parts. The yields from PEGDA-based resin and Teflon-coated
Pro3dure GR-1 were below the detection limit (Figure 34). . . . . . . . . . . . . . . . . . . 80
34 Transcription using Water with leachate. The gel image shows the transcription outcome
with leachates collected from different materials. Electrophoresis Buffer: 1X TBE. . . . . . 80
xi
Abstract
In recent years, microfluidic systems have found broad applications in biopharmaceutical processes. The
unique advantages of microfluidics include precise fluid control, enhanced mixing, small footprint, and
compatibility with fluid control equipment for automation. These qualities are particularly appealing to
the biopharmaceutical community because of the inherently sophisticated biochemical procedures that
requires precise control over precious analytical samples and synthetic precursors. For instance, well-
controlled microscale mixing has attracted widespread interest for nanodrug fabrication. Despite having
compelling potential for pharmaceutical applications, the industry has been slow in adopting microfluidics.
This is largely due to the lengthy fabrication process and poor fluidic interfaces of traditional microfluidic
devices. To bridge the gap between microfluidics research and industrial applications, the microfluidics
community has taken to additive manufacturing (i.e. 3D printing) for microfluidic device fabrication be-
cause of its unprecedented ease of use, rapid prototyping, 3D capability for user-friendly fluidic interfaces,
and low cost. Further exploration in utilizing 3D-printed microfluidic devices biopharmaceutical applica-
tions is critical to disseminate microfluidics beyond small-scale research settings.
This dissertation begins with a background on microfluidic technology, its applications in pharmaceu-
tical development, and 3D printing for microfluidic device fabrication. The research work in the following
chapters focus on microfluidic applications in nanodrug fabrication and affinity reagent generation. In
chapter 2, microfluidic liposome production and concurrent loading of drug simulants is introduced. The
xii
precise control and enhanced mixing in microfluidics enable fine tuning of liposome sizes and drug en-
capsulation efficiencies by simply manipulating the flow parameters. In chapter 3, we introduce a novel,
3D-printed hydrodynamic flow focusing device that employs an elegant 4-way sheath flow channel and
a downstream staggered herringbone mixer for producing mRNA-encapsulated lipid nanoparticles. At
high throughput, the device produces size-limited lipid nanoparticles with high mRNA encapsulation ef-
ficiencies owing to its superior fluid focusing and rapid mixing capabilities. In chapter 4, a 3D-printed
microfluidic device that enables kinetic off-rate selection for affinity reagent development is presented. By
using a continuous flow technique, kinetic off-rate selection is achieved with improved selectivity for high
affinity ligands. This work encourages the development of microfluidic platforms for automated affinity
ligand screening. To explore the utility of 3D-printed microfluidics for biochemical reactions, we eval-
uated commercially-available, photocurable resins for stereolithography printing for their compatibility
with fundamentalinvitro enzymatic reactions, including PCR, transcription, translation, and reverse tran-
scription. We found that most resins demonstrated low enzyme compatibility because of surface adsorption
and toxic leachates that diffuse from the printed parts during the reaction. The closing chapter concludes
this dissertation and provides future perspectives on 3D-printed microfluidics for the biopharmaceutical
research community.
xiii
Chapter1
Introduction
1.1 MicrofluidicsforBiopharmaceuticalApplications
Microfluidics technology processes fluids that are geometrically constrained within a micrometer-scaled
channel, typically within 10-900µ m. The distinct behavior of fluids at this scale allows precise control of
small volumes of fluids and convenient manipulation of fluid interfaces. In microfluidic channels, viscous
forces dominate inertial forces and diffusive transport dominates advective transport, so the kinetics of
the system are more predictable and controllable than fluids at macroscale. Additionally, the dominating
surface tension, interfacial tension, and capillary forces over gravitational forces in microfluidic channels
can be conveniently leveraged for fluid manipulation. These appealing features of microfluidics, along with
its inherently small processing volume, make the technology particularly suitable for biopharmaceutical
processes where materials are precious and/or fluid control at the micrometer-scale is desired[42].
In recent years, microfluidic-based biopharmaceutical technologies, especially for drug synthesis and
discovery, have been developed as improvements to their conventional counterparts. Lipid-based nanopar-
ticles have been used to deliver therapeutic agents or serve as labels for various biomolecule detection
assays. For use as nanomedicine, their sizes typically range from 30 - 200 nm for efficient cellular de-
livery[38, 179, 147, 31]. Because of the stringent particle size requirement, microfluidic approaches to
synthesize lipid-based nanoparticles have emerged. Precise control over localized fluid mixing enables the
1
generation of size-limited nanoparticles. Aside from pharmaceutical synthesis, microfluidics have found
applications in targeted drug delivery developments. Targeted drug delivery differs from conventional
drug delivery in that the therapeutic substances are administered using nanoparticles as drug carriers in
a manner that concentrates the therapeutic substances in particular parts of the body relative to others.
This technique offers increased therapeutic efficacy with minimal side effects, which is pivotal in many
cancer therapies. The pharmaceutical development processes for targeted drug delivery often depletes
precious analytical samples, which means that microfluidic approaches that miniaturize biochemical pro-
cedures to reduce materials consumption are beneficial. From pharmaceutical development to production,
microfluidic technology has been widely explored for its unique advantages of precise fluid manipulation,
well-controlled microenvironment, low energy consumption, and reduced materials consumption[138, 33,
82].
1.1.1 MicrofluidicSynthesisinNanomedicine
The small sizes of nanoparticle delivery systems offer prolonged circulation time and enhanced cellular
uptake compared to large particle delivery systems and molecular therapeutics[125, 126]. By encapsu-
lating small therapeutic molecules in nanocarriers, the drug molecules can be protected and directed to
areas for enhanced efficacy and reduced toxicity to unwanted areas in contrast with small molecule ther-
apeutics[126]. Over the last few decades, more than 100 nanoparticle drug delivery systems have been
approved by the U.S. Food and Drug Administration (FDA) or are currently in clinical trials[1]. Recently,
mRNA vaccines that employ lipid nanoparticles for cellular delivery have been administered for protection
against coronavirus disease 2019 (COVID-19) worldwide[61]. These vaccines are produced using microflu-
idic technology, in which the mRNA-loaded lipid nanoparticles are synthesized by rapid microscale mixing.
The controlled, localized mixing within the micro-environment sets microfluidics approaches apart from
conventional nanomedicine synthesis techniques[176]. Conventional techniques, such as breaking down
2
bulk products or large particles, suffer high polydispersity and large batch-to-batch variability due to the
lack of precise control over the nanoparticle formation process[111]. For drug delivery applications, this
is problematic as the size of the nanoparticles has substantial impact on the delivery efficiency and release
profile of the drug[47, 37, 7].
In microfluidic synthesis techniques, homogeneous nanoparticles are generated by microscale mixing
of two fluid phases[176]. For nanomedicines that are formed by self-assembly of amphiphilic molecules,
such as liposomes, particle size is directed by the diffusion of the molecules across a liquid-liquid in-
terface. As the flow in microfluidic devices is usually laminar, this diffusion-dominated mixing is well-
controlled[42]. This simple mixing mechanism permits direct control of the mixing time by manipulating
the microfluidic structure and flow rates, and therefore enables tunable particle characteristics. While dif-
fusion of the amphiphilic molecules determines particle size, the drug loading efficiency (i.e. the amount of
drug that gets encapsulated within the nanoparticles relative the the amount of drug in the total mixture)
relies on rapid convective mixing of the two liquids[102]. For enhanced mixing, various microfluidic struc-
tures have been studied, including T-junctions[177], V-junctions[56], capillaries[44], coaxial tubes[93], and
herringbone mixers[174]. In general, nanoparticles fabricated using microfluidic techniques are more uni-
form in size and have higher drug loading efficiency compared to the ones produced by bulk mixing[176].
Microfluidic nanoparticle synthesis allows fine tuning of the synthesis parameters (i.e. flow rates, tem-
perature, pH) with high reproducibility, low reagent consumption, and reduced synthesis time[176]. For
high-throughput nanoparticle production, microfluidic systems can be easily scaled-up through paralleliz-
ing multiple microchannels[167]. These features of microfluidic systems provide compelling advantages
over other nanoparticle synthesis techniques for drug delivery.
3
1.1.2 MicrofluidicsforAffinityReagentDevelopment
For next-generation diagnostics and therapeutics, newly identified potential molecular targets require fur-
ther studies in order to be utilized. Affinity reagents that bind specifically and strongly to these targets
are essential tools for cancer target research and their development for therapeutic and diagnostic usage.
As many molecular targets are being identified thanks to the advancements in cancer genetics, the devel-
opment of cancer diagnostics and therapeutics depends on our ability to generate target-specific affinity
reagents. Developed by César Milstein and Georges J. F. Köhler in 1975, hybridoma technology (mono-
clonal antibodies; mAbs) remains the primary reagent generation method[86]. The reagents generated
by hybridoma technologies are developed from antibodies produced by immortalized B cells, which are
fused mouse immunized B cell and myeloma cells[86]. The development typically takes six months or
longer, require large amounts of precious target (50-100 mg), and the use of host animals prohibits ap-
plications to toxic antigens and requires further humanization to be used as therapeutic reagents[20]. To
address these challenges, numerous molecular display technologies have been developed to screen affinity
reagents, many of which have been integrated and improved with microfluidic techniques.
Molecular display technologies tailor molecules (proteins, polypeptides, nucleic acids and others) for
high affinity and specificity against biochemical targets. The key strategy of molecular display technologies
is the physical coupling of phenotype to genotype (e.g. protein to its encoding nucleic acid), which en-
ables the recovery of the information encoding a molecule sequence for amplification after target-directed
selection. Through repeated screening and amplification, high affinity molecules are enriched and iden-
tified from a combinatorial library. The most common technique to associate phenotypes to genotypes
is by displaying peptides and polypeptides on filamentous phage (phage display)[150], while bacteria[49]
or yeast[19] have also been used to express proteins on their surfaces. Alternatively, cell-free selection
technologies that attach peptides and proteins directly to their encoding genes via physical or chemical
bonds have also been explored extensively, such as ribosome display and mRNA display. Although all
4
the above-mentioned methods have been sought out as powerful tools to screen for affinity reagent, their
development is still time-consuming and labor intensive.
Recent advancements in molecular display technologies involve using microfluidics to automate the
biochemical processes for optimization and high-throughput screening[138]. Microfluidic approaches to
simplify the tedious biochemical processes have emerged to improve the throughput, enhance the pro-
cess robustness, minimize the consumption of target material, and increase reproducibility of molecular
display technologies. Molecular display technologies integrated with microfluidic systems require con-
siderably smaller quantities of samples and reagents, which can reduce the cost to develop reagents since
target material is often precious. For instance, microfluidic phage display resulted in a fifteen-fold time
reduction and a ten-fold lower reagent volume consumption per target compared to its large-scale counter-
part[34]. Aside from increasing time-efficiency, integrated microfluidic systems for automated molecular
display technologies demonstrated higher reproducibility owing to the precise control of reaction condi-
tions (e.g., pH, temperature, and shearing forces) and enabled high-throughput screening by parallelizing
the microchannels through multiplexing[138, 33].
Although biochemical laboratory automation is currently dominated by fluid handling robots, inte-
grated microfluidic systems have remarkable scalability and considerably lower equipment cost[138]. It is
envisioned that these microfluidic technologies will pave the way to affordable, compact, and automatic
platforms for rapid affinity reagent development.
1.2 Towards3DPrintedMicrofluidics
1.2.1 TheRiseof3DPrintingforMicrofluidics
Microfluidic devices capable of performing biochemical processes on the micrometer scale have revolution-
ized chemical and biological research[136, 12, 168]. Early microfluidic devices were fabricated with silicon
5
and glass, but the fabrication processes were cumbersome[158, 58, 104, 73]. Elastomer micromolding,
which involves molding the elastomer polydimethylsiloxane (PDMS), was first described by the White-
sides group in 1998 and has since became the dominant fabrication method for microfluidic devices[170].
Elastomer micromolding typically involves forming open microfluidic channels on a PDMS film by curing
PDMS on a mold that consists of the complementary microfluidic structures. After curing the PDMS film,
the open channels are closed with a glass slide or another piece of PDMS via plasma-activated bonding.
The widespread use of PDMS for microfluidic devices is due to its affordability and attractive physico-
chemical properties that are well suited for biomedical applications[16]. PDMS is easy to cure, chemically
inert, optically transparent, flexible to surface modifications, gas permeable, and biocompatible[182, 145,
16].
Although PDMS-based microfluidic devices have shown success, several challenges emerged as a result
of the growing demand for increasingly sophisticated microfluidic designs and calls to translate microflu-
idics technologies to industrial applications[114]. Most PDMS-based microfluidic devices incorporate del-
icate fluidic interfaces that are prone to leaks and awkward to connect, making the devices incompatible
for industry usage[83, 25, 16]. The control systems of the microfluidic components require engineering
expertise for operation, which limits the use of these systems to trained individuals. Further, the typical fab-
rication process for the complementary mold is labor-intensive[110], and its alternative high-throughput
molding techniques (e.g. injection molding) are costly and impractical for prototyping. These challenges
have motivated the field to take a new turn towards 3D printing for microfluidic device fabrication[16].
3D printing, also known as additive manufacturing, is a fabrication technique that constructs a three-
dimensional object through consecutive layers of material patterning[54]. For prototyping, 3D models are
designed using computer aided design (CAD) software where each surface of the 3D model is defined. A
3D printer converts the model of the desired object into a sequence of two-dimensional horizontal cross
sections that are then printed additively. 3D printing enables affordable, single-step, and rapid small-scale
6
production of three-dimensional structures, which was previously unattainable in other manufacturing
techniques. While it is not intended to replace high-throughput manufacturing technologies, 3D print-
ing has revolutionized the prototyping workflow in various fields from research to consumer product
development[16, 141, 119, 165, 166]. Recent interests in 3D printing for rapid prototyping have echoed
within the microfluidic community. The transition from traditional elastomer micromolding to 3D print-
ing opens the possibility of affordable, assembly-free fabrication of integrated microfluidic devices with
three-dimensional capabilities. Aside from the superior manufacturability, 3D microfluidic devices can
be printed with interfaces that are compatible with commercially available fluidic connectors (e.g. 1/4-28
metric threaded connectors from IDEX Health and Science). For microfluidic automation, the user-friendly
interfaces allow facile implementation of programmable fluid handling equipment, such as syringe pumps
and solenoid valves, that are affordable and readily available in many scientific research laboratories.
1.2.2 3DPrintingTechnologiesforMicrofluidics
In the past decade, the dimensional resolutions of 3D printing have reached the micron-scale[105]. The
technology has thus found applications in biomedical research to replace traditional microfabrication tech-
niques that are costly and labor-intensive[16]. Among various 3D printing techniques, inkjet 3D print-
ing (i3DP), fused deposition modeling (FDM), and stereolithography (SLA) are most relevant to microflu-
idics[166].
i3DP was the first 3D printing technology to be used for microfluidic device fabrication[109]. In i3DP,
cross-sections of a three-dimensional objects are created using inkjet technology. Operations of inkjet
can be continuous or drop on demand (DoD)[146]. In DoD operations, a pulse is generated to push a
droplet of materials from the nozzle as opposed to continuous dispensing[35]. For microfabrication, DoD
is more favorable for its higher resolution[35]. In powder-based i3DP, a layer of cross-section is formed by
uniformly depositing a thin layer of powder on the build platform on which droplets of polymeric adhesive
7
solutions are jetted prior to forming a subsequent layer[35]. Alternatively, photopolymer-based i3DP uses
an array of inkjet nozzles to deposit droplets of photopolymers by layers[35]. For microfluidic device
fabrication, a major impediment of i3DP is the need to print support structures within the microfluidic
channels[166]. Although i3DP allows for printing channels down to 25µ m, the support structures are
difficult to remove afterwards[166]. Additionally, the printers need to be used regularly to prevent clogging
and are typically 10–100 times more expensive than STL and FDM printers[166]. Despite these limitations,
i3DP appeals to the microfluidics community due to its capability to print multiple materials in one step
and unmatched reliability[63].
Developed by Scott Crump from Stratasys in 1990, FDM is one of the most widely used manufacturing
technologies for rapid prototyping[32]. In FDM, thermoplastic filaments are melted and extruded onto a
print stage to form a patterned layer[127]. FDM printers consist of a moving extruder with a fine nozzle
tip that is temperature-controlled to melt the thermoplastic, which is then solidified in desired areas as the
extruder traces the design of each cross-sectional layer. Therefore, the printing resolution depends on the
vertical dimensional accuracy of the extruder and is ultimately restricted by the size of the nozzle tip[172,
63]. For microfluidic device fabrication, FDM is limited to printing microchannels that are over 100 µ m in
dimension with rough channel surfaces[166]. Despite its inherent limitations in dimensional accuracy and
surface texture, FDM offers some advantages over other 3D printing techniques. These printers do not re-
quire support structures to be printed within the microchannel and some are capable of multiple-material
printing, which enables one-step manufacturing for integrated microfluidic device[46]. Further, FDM en-
ables 3D printing of a wide range of materials that are compatible with mass production techniques[112,
41, 27, 181]. These advantages, along with its affordability, show great promises for microfluidic device
fabrication as the resolution continues to be improved.
Chuck Hull introduced stereolithograhpy (SLA) in the 1980s and defined it as “a method and apparatus
for making solid objects by successively printing thin layers of the ultraviolet curable material one on top
8
of the other”[68]. In SLA, the additive cross-sectional layers of an object are photopolymerized in a bath
of liquid resin using a scanning laser or a digital light projector. The object is polymerized onto a moving
print bed that travels away from the light source after each layer is formed. A major advantage of SLA
for microfluidic device fabrication is its superior spatial resolution and smoother surface texture compared
to above-mentioned methods[16]. The resolution of SLA depends on the resolution of the light source
and the properties of the resin after polymerization. For printing microflulidic devices, over-curing of the
resins within microfluidic channels limits the print resolution to larger, 100-500 µ m channel dimensions.
Recently, Chen and colleages solved this issue by implementing in-situ transfer vat photopolymerization
and demonstrated printing of 10µ m-sized microfluidic channels[173]. For ultra-high resolution printing,
multi-photon optics can be employed to tightly focus the high-intensity pulsed laser beams to femtoliter
volumes[16, 81, 142, 88, 80]. Although SLA requires draining uncured resins from the microfluidic channels
after printing, the removal process is easier than that of i3DP since the resins are in liquid form[166]. These
advantages make SLA particularly suitable for microfluidic device fabrication. However, unlike FDM or
i3DP, SLA only prints with one material at a time and the material selection is intrinsically limited to
photopolymerizable resins[166].
The rise of 3D printing owes to its unprecedented ease in fabricating three-dimensional objects at low
cost. Recent growth in affordable printers and 3D printing services makes 3D printing increasingly acces-
sible. Despite early enthusiasm for 3D printing microfluidics, 3D printing technologies have yet to replace
elastomer micromolding for microfluidic device fabrication because of current shortfalls. i3DP and SLA
can print small microchannels, but i3DP requires removal of support material and SLA requires drainage of
uncured resins. Devices fabricated with photopolymer-based i3DP and SLA have limited material choice.
While FDM can print microfluidic devices with a wide range of thermopolymers, the obtainable resolution
is quite large by microfluidic standards and the resulting channel surfaces are rough. These limitations
9
have prompted the continuous development in 3D printing technologies, which is fundamental to the
success of 3D printing microfluidics.
10
Chapter2
MicrofluidicLiposomeProductionandConcurrentLoadingofDrug
Simulants
Note: This chapter has been published as a journal article. The full reference is: Wan-Zhen Sophie Lin
and Noah Malmstadt. “Liposome production and concurrent loading of drug simulants by microfluidic
hydrodynamic focusing”. In: European Biophysics Journal 48.6 (Sept. 2019), pp. 549–558. issn: 0175-7571,
1432-1017. doi: 10/gqjqcx. url: http://link.springer.com/10.1007/s00249-019-01383-2 (visited on
07/12/2022)
2.1 Abstract
Liposomes are spherical vesicles enclosed by phospholipid bilayers. Nanoscale liposomes are widely em-
ployed for drug delivery in the pharmaceutical industry. In this study, nanoscale liposomes are fabricated
using the microfluidic hydrodynamic focusing (MHF) approach, and the effects of flow rate ratio (FRR)
on liposome size and drug loading efficiency are studied. Fluorescein isothiocyanate modified dextran is
used as a hydrophilic drug simulant and Nile red is used as a hydrophobic drug simulant. The experiment
results show that hydrophilic drug simulant loading efficiency increases as FRR increases and eventually
plateaues at around 90% loading efficiency. The hydrophobic drug simulant loading efficiency and FRR
have a positive linear correlation when FRR varies from 10 to 50. Concurrent loading of both hydrophilic
11
and hydrophobic drug simulants maintains the same loading efficiencies as those of loading each drug
simulant alone. A negative correlation between liposome size and FRR is also confirmed. Unloaded lipo-
somes and hydrophilic drug-loaded liposomes are of the same sizes, and are smaller than the ones loaded
with the hydrophobic drug simulants alone or combined. The results suggest tunable liposome size and
drug loading efficiency with the MHF technique. This provides evidence to encourage further studies of
microfluidic liposome fabrication in the pharmaceutical industry.
2.2 Motivation
Liposomes are spherical shells consisting of phospholipid bilayers that enclose small volumes of aqueous
solution. The formation of liposomes is driven by the hydrophobic effect which minimizes the interactions
between hydrophobic and hydrophilic molecules by spatial organization. Liposomes were first discovered
in 1965 and soon applied to drug delivery owing to their biocompatibility[8, 53]. Today, liposomes are
widely employed to deliver biologically active compounds, such as proteins, enzymes, hormones, DNA vec-
tors, anti-cancer drugs, and antimicrobial agents into cells[159]. Liposomes can encapsulate hydrophilic
pharmaceutical agents in their internal aqueous compartments and entrap hydrophobic pharmaceutical
agents in their phospholipid bilayers[106]. One of the biggest challenges in liposome pharmaceutical ap-
plications is the fast detection and clearance of liposomes by the renal system. With larger curvature,
smaller liposomes inversely affect the number of recognition sites and thus correlate to lower clearance
rate[97]. Therefore, it is important to produce nanoscale liposomes of controlled size for liposome-based
drug delivery applications[22, 64].
Current liposome production in the pharmaceutical industry uses batch techniques, in which lipids
are hydrated in aqueous buffers to form liposomes in bulk[64]. Liposome formation in bulk, such as thin
film hydration and reverse-phase evaporation, often results in large, polydisperse, and multilamellar li-
posomes that require additional steps to yield a uniform liposome size[64, 53, 98, 175, 153, 129, 155, 10,
12
108, 57, 21]. Unlike traditional methods, liposome production by microfluidics allows continuous produc-
tion, scale-up production by parallelization, controlled liposome size, reduced production time and cost,
reduced solvent consumption, and uses less toxic solvents[21, 9, 103]. Microfluidic liposome fabrication
typically involves rapid diffusive mixing of two or more inlets containing a lipid mixture in alcohol and
an aqueous buffer. When the alcohol concentration decreases below a critical concentration, lipids be-
come insoluble and spontaneously self-assemble into liposomes. Nanoscale liposomes are formed within
a confined microenvironment[22, 64]. Microfluidic hydrodynamic focusing (MHF) can produce uniformly
dispersed nanoscale liposomes, allowing for liposome size control via adjustments to inlet volumetric flow
rates.[22, 64, 103, 74] The ability to control liposome size without additional steps allows instant liposome
preparation for a variety of research and pharmaceutical applications[64, 74].
In this chapter, we demonstrate tunable drug loading efficiencies and liposome sizes using the MHF
technique. In the MHF technique, a stream of lipid in alcohol solution is intersected and sheathed by
two lateral streams of hydration buffer[22, 64, 103, 74]. The confined microscale channel facilitates lami-
nar flow and diffusion between the phases. The diffusion initiates liposome formation by decreasing the
surrounding alcohol concentration to below the critical alcohol concentration[74]. The liposome forma-
tion mechanism in this technique was described previously[74]. Briefly, the liposomes experienced initial
self-assembly at the alcohol-buffer interface, disassembly due to the diffusion of the alcohol phase, and
eventual reassembly as the two phases blend. According to the previous study, increased alcohol stream
width and shorter diffusion period were observed at lower inlet flow rate ratio of the hydration buffer
over the alcohol stream[74]. Therefore, at lower inlet flow rate ratio, the magnitude of the disassemble-
reassemble phenomenon increased[74]. The magnitude of the phenomenon positively affected the size of
the reassembled liposomes, as higher input flow rate ratio resulted in smaller liposomes[74]. When the
FRR was further increased, smaller changes in the stream width and diffusion period were observed and
13
the liposome size eventually decreased to a limit[74]. The minimum of obtainable liposome size depended
on the dimensions and geometry of the microfluidic device[74].
For analysis purposes, fluorescein isothiocyanate modified dextran (FITC-dextran) was used as a hy-
drophilic drug and Nile red was used as a hydrophobic drug. Their fluorescence intensities were mea-
sured using a microplate reader and liposome sizes were detected by dynamic light scattering (DLS).
FITC-dextran and Nile red concentrations were calculated based on their calibrated fluorescence inten-
sity. Loading efficiency was defined as the concentration of drugs loaded onto liposomes divided by the
drug concentration of the collected sample. Total flow rate (TFR) was defined as the sum of the input flow
rates of the hydration buffer and the lipid mixture, and flow rate ratio (FRR) was defined as the ratio of the
input hydration buffer flow rate over the input lipid mixture flow rate. This study was designed to provide
experimental evidence to strengthen the feasibility of the MHF technique, particularly for industrial ap-
plications such as in-vivo studies and on-demand liposome preparations. The ease of single-step liposome
fabrication and concurrent drug loading could encourage further research on an industrial-scale liposome
production by MHF.
2.3 MaterialsandMethods
2.3.1 DeviceFabrication
Positive microfluidic channels were patterned on silicon wafers by photolithography using SU-8 50 neg-
ative tone photoresists (MicroChem Corp., Newton, MA). PDMS was placed onto silicon wafers to form
negative channels. A hole was punched at each fluidic access point using a biopsy puncher (outer diameter
1.5 mm) and polytetrafluorethylene tubing (1/16” outer diameter and 1/32” inner diameter) was inserted.
Each PDMS device was attached to a thin piece of glass after corona treatment (BD-20 Corona Treater,
Electro-Technic Products Inc., Chicago, IL). The resulting channels had rectangular cross sections of 50µ m
14
in depth and either 60µ m or 80µ m in width (Figure 1). The inlet tubing was connected to syringes and the
input flow rates were controlled by syringe pumps (Genie Touch, Kent Scientific, Torrington, CT).
(A) (B)
Figure1: The microfluidic hydrodynamic focusing device. (A) Hydration buffer solution and lipid solution
in IPA entered the device through fluidic access points a and b, respectively. The liposome product was
collected from fluidic access point c. (B) The height of the channel was 50 µm, and the width was 60 or
80µm.
2.3.2 LipidMixtureandHydrationBufferPreparation
Dimyristoylphosphatidylcholine (DMPC) (Avanti Polar Lipids Inc.), cholesterol (Avanti Polar Lipids Inc.),
and dihexadecyl phosphate (DCP) (Sigma-Aldrich) were dissolved in chloroform at a 5:4:1 molecular ra-
tio[74]. The solution was placed in a vacuum desiccator at room temperature to evaporate the solvent
to form a dry lipid film[74]. The dry lipid film was then dissolved in dry isopropyl alcohol (IPA) to form
a solution (5mM total lipid concentration)[74]. Diluted phosphate buffered saline (PBS) solution (0.01M
NaCl, pH = 7.40, Sigma-Aldrich) was used as the hydration buffer[74].
2.3.3 LiposomePreparation
The hydration buffer entered the channel through fluidic access point a, while the lipid solution entered
the channel through fluidic access point b (Figure 1). The two solutions converged at the channel’s joint
where liposome formation begun. The resulting liposomes were collected from access point c (Figure 1).
15
Flow rate ratio (FRR) was defined as the flow rate ratio of hydration buffer flow rate to lipid solution flow
rate; it varied from 10 to 50 while maintaining an inlet lipid mixture flow rate of 3 µ L/min. Total flow
rate (TFR) was defined as the total flow rate of hydration buffer flow and the lipid solution; it varied from
62µ L/min to 186µ L/min and maintained a constant FRR of 30. The radii of the liposomes were measured
by Dynamic Light Scattering (DLS) (Dynapro Titan, Wyatt Technology Corporation, Santa Barbara, CA).
2.3.4 HydrophilicDrugLoading
4kDa, 20kDa, and 40kDa fluorescein isothiocyanate modified dextran (FITC-dextran) (Sigma-Aldrich) were
used as the hydrophilic drug simulants. Each was dissolved in a hydration buffer at 4.0 µ g/mL. FITC-
dextran concentrations were measured by a Microplate Reader (480nm /520nm, Synergy H1 Hybrid Multi-
Mode Microplate Reader, BioTek, Winooski, VT) and the liposome radius was measured by DLS. The li-
posomes were filtered out using Amicon Ultra-4 centrifugal filter units (MWCO 100 kDa) (Millipore, Ger-
many) at 3220µ g for 1 hour at room temperature to collect the filtrate. The drug loading efficiency was
calculated by equation 2.1, where the FITC-dextran concentration in the filtrate was the concentration
of the collected filtrate after ultracentrifugation and the inlet FITC-dextran concentration was the drug
concentration of the collected liposome product.
%loading efficiency =

1− FITC-dextran Concentration in Filtrate
Inlet FITC-dextran Concentration

× 100% (2.1)
2.3.5 FITC-dextranDiffusivityCalculation
The hydrodynamic radii of 4kDa, 20kDa, and 40kDa FITC-dextran were obtained from the product infor-
mation provided by the manufacturer (Sigma-Aldrich). The intrinsic viscosity is defined as the solute’s
viscosity contribution. It was calculated using the hydrodynamic radius as shown in equation 2.2, where
16
[η ] was the intrinsic viscosity,R
h
was the molecule hydrodynamic radius, N was Avogadro’s number, and
M was the molecular weight.
[η ] =
10R
3
h
π N
3M
(2.2)
The relationship between intrinsic viscosity and solution viscosity was as shown in equation 2.3, where
η was the solution viscosity, η 0
was the solvent viscosity, and ϕ was the solution concentration. Since
the FITC-dextran concentrations were low, equation 2.3 was simplified and rearranged to equation 2.4 to
calculate the solution viscosity.
[η ] = lim
ϕ →0
η − η 0
η 0
ϕ (2.3)
η =η 0
(ϕ [η ]+1) (2.4)
The diffusivities were calculated by the Stokes-Einstein relation as shown in equation 2.5, where D was
the diffusivity, k
B
was the Boltzmann’s constant, and T was the absolute temperature.
D =
k
B
T
6πηR
h
(2.5)
2.3.6 HydrationBufferRecycling
Liposomes loaded with FITC-dextran were filtered out using centrifugal filters at 3220 µ g for 1 hour at
room temperature. The collected filtrate was a mixture of unloaded FITC-dextran, hydration buffer, and a
small amount of IPA. The solvent was used as the hydration buffer for another trial of liposome production.
Hydration buffer recycling was repeated three times, and the FITC-dextran loading efficiency and mean
liposome size of each trial were analyzed. Three separate experiments were carried out at FRR of 20, 30, and
50 at a constant lipid mixture input flow rate of 3 µ L/min. The effect of IPA contamination in hydration
buffer on liposome production was assessed by varying the IPA concentration from 0% to 9.09%. The
effect of FITC-dextran concentration in hydration buffer on liposome production was assessed by varying
17
FITC-dextran concentration from 1µ g/mL to 4µ g/mL. FITC-dextran concentrations were measured by the
Microplate Reader (480 nm/520nm), FITC-dextran loading efficiencies were calculated by equation 2.1, and
liposome sizes were determined by DLS.
2.3.7 HydrophobicDrugLoading
Nile red (Thermo Fisher Scientific) was used as the hydrophobic drug simulant and was dissolved in the
lipid in IPA solution at 108µ g/mL. The entrapped Nile red concentrations were measured by the Microplate
Reader (552 nm/636nm), and liposome radius was measured by DLS. Since the fluorescence of Nile red was
quenched in aqueous solution, the fluorescence of the collected product reflected the concentration of Nile
red trapped within the membrane. Nile red loading efficiency was calculated by equation 2.6, where the
concentration of entrapped Nile red was the concentration determined by the fluorescence intensity of
the collected sample, and the inlet Nile red concentration was the calculated Nile red concentration in the
solution based on the inlet concentration and the final volume of the product.
%loading efficiency =
Concentration of Entrapped Nile red
Inlet Nile red Concentration
× 100% (2.6)
2.3.8 HydrophilicandHydrophobicDrugLoading
4kDa FITC-dextran and Nile red were used as the hydrophilic and hydrophobic drug simulants, respec-
tively. 4kDa FITC-dextran was dissolved in the hydration buffer at 4 µ g/mL, and Nile red was dissolved
in the lipid solution at 108µ g/mL. Loading efficiencies of FITC-dextran and Nile red were obtained in the
same manner as described above, and liposome radii were measured by DLS.
18
2.3.9 StatisticalAnalysis
Analysis of variance (ANOVA) was used to determine whether or not the correlations of the experiment
results were significant. Correlations within a 95% confidence level (p < 0.05) were considered significant.
Each size distribution was monodisperse with a polydispersity less than 15%. Each experiment for loading
efficiency was repeated at least three times to obtain a standard deviation shown as the error bars in the
figures. The error bars in the radius data represented the standard deviation obtained from the %polydis-
persity measured by DLS. The error bar for diffusivity was obtained by error propagation.
2.4 ResultsandDiscussion
Hydrophilic Drug Loading. FITC-dextran loading efficiency increased as the flow rate ratio (FRR) in-
creased and eventually plateaued at around 90% (Figure 2(A)). The phenomenon can be described by a
theory presented in a previous study[74]. Briefly, liposomes formed by the MHF method experience ini-
tial rapid liposome self-assembly at the alcohol-buffer interface, followed by partial disassembly and re-
assembly[74]. At the solvent-buffer interface, the alcohol concentration is much lower than the critical
alcohol concentration, triggering rapid liposome self-assembly[74]. Due to their low diffusivity, the ini-
tially formed liposomes travel along the stream, while more alcohol diffuses into the buffer and induces an
alcohol concentration gradient normal to the stream[74]. The initially formed liposomes disassemble into
lipids or sections of lipid bilayer when the surrounding alcohol concentration is higher than the critical
concentration. Eventually the liposomes reassemble when the surrounding alcohol concentration is just
below the critical concentration as the two phases mix[74]. At lower FRR, the alcohol stream width was
wider, which increased the magnitude of the disassembly-reassembly effect. When liposomes disassem-
bled due to the alcohol concentration gradient, the initially encapsulated FITC-dextran was released to
the surrounding environment. Later, liposomes reassembled when the surrounding alcohol concentration
19
was just below the critical concentration and encapsulated a small amount of FITC-dextran due to the
FITC-dextran concentration gradient. At higher FRR, liposomes reassembled at higher surrounding FITC-
dextran concentrations due to a narrower alcohol stream width and shorter diffusion period. Therefore,
fewer liposomes experienced the disassembly-reassembly effect. When the FRR was increased to a point
that the surrounding alcohol concentration never returned above the critical concentration, the initially
formed liposomes never disassembled and the system obtained an approximately 90% loading efficiency.
(A) (B)
(C) (D)
Figure 2: FITC-dextran loading. (A) 4kDa, 20kDa, and 40kDa FITC-dextran loading efficiencies corre-
sponding to different FRRs. The lipid mixture inlet flow rate was controlled at 3 µL/min. (B) Mean liposome
radii corresponding to different IPA concentrations in the hydration buffer inlet. (C) FITC-dextran loading
efficiency at varying TFR and a constant FRR of 30. (D) Mean liposome radii at varying TFR and a constant
FRR of 30.
20
The results of loading 4kDa, 20kDa, and 40kDa FITC-dextran were compared. At FRRs lower than
those achieved at plateau encapsulation, 40kDa FITC-dextran loading efficiency was the highest, and 4kDa
FITC-dextran loading efficiency was the lowest (Figure 2(A)). The effect of molecular weights on loading
efficiency was statistically significant (ANOVA, 95% confidence level). To delve into this phenomenon, the
FITC-dextran diffusivities of different molecular weights were approximated (Figure 3(A)). FITC-dextran
diffusivity and loading efficiency had a negative correlation (Figure 3(B)). This suggests that larger FITC-
dextran molecules have lower diffusivity, and therefore maintained a higher concentration at the stream
interface during liposome reassembly (Figure 3).
(A) (B)
Figure 3: FITC-dextran diffusivity and loading efficiency. (A) 4kDa, 20kDa, and 40kDa FITC-dextran
diffusivities were approximated. (B) FITC-dextran loading efficiency corresponding to diffusivity. The
error bars presented the uncertainties of the data obtained by error propagation.
Liposome radius decreased as FRR increased (Figure 2(B)); consistent with the results of previous
studies[74, 184]. This was due to the attenuation of the disassembly-reassembly phenomenon at higher
FRR[74]. Liposomes loaded with FITC-dextran had similar sizes than those without drug simulants, and
FITC-dextran molecular weight had no effect on liposome size; liposome size was stable over 20 hours
(Figure 2(B)). This indicates that the amount of fluid trapped inside a liposome was not influenced by
FITC-dextran loading or FITC-dextran molecular weight. At a constant FRR of 30, no correlation was
found between TFR and FITC-dextran loading efficiency or liposome size (Figure 2(C)).
21
HydrationBufferRecycling. Due to the high FRR, a considerable amount of hydrophilic drug was
not encapsulated in liposomes and remained in the filtrate of the sample despite the high loading effi-
ciencies. Therefore, the potential of reusing the filtrate of the sample as the hydration buffer to fabricate
liposomes was investigated. The results showed that liposome radius increased and FITC-dextran loading
efficiency decreased each time the solvent was reused as the hydration buffer (Figure 4).
(A) (B)
Figure4: Reusing product filtrate as the hydration buffer for liposome fabrication. Three trials were done
at FRR of 20, 30, and 50. The lipid mixture inlet flow rate was controlled at 3 µL/min while the FRR varied.
(A) FITC-dextran loading efficiency corresponding to the number of times the collected filtrate was reused
as a hydration buffer inlet. (B) Mean liposome radii corresponding to the number of times the filtrate was
reused as a hydration buffer inlet.
Since the solvent had lower FITC-dextran concentration and higher IPA concentration each time it
was reused, the effects of FITC-dextran concentration and IPA contamination of the hydration buffer were
investigated to further understand the cause of the phenomenon. As shown in Figure 5, the decrease in
FITC-dextran loading efficiency and increase in liposome sizes were most likely the results of IPA contam-
ination, since the variation in FITC-dextran concentration showed no effect. One possible explanation is
that the IPA contamination in the buffer stream extended the period during which the surrounding IPA
concentration was higher than the critical concentration, and thus intensifies the disassembly-reassembly
effect. The FITC-dextran diffusion rate toward the center of the stream increased due to the contamination
22
and thereby resulted in lower surrounding FITC-dextran concentration during liposome self-reassembly.
The reassembly was less rapid and therefore resulted in larger liposomes (Figure 5(B)).
(A) (B)
(C) (D)
Figure5: FITC-dextran loading at varying %IPA and FITC-dextran concentration in the hydration buffer.
The lipid mixture inlet flow rate was controlled at 3 µL/min. (A) FITC-dextran loading efficiencies cor-
responding to different IPA concentrations in the hydration buffer inlet. (B) Mean liposome radii corre-
sponding to different IPA concentrations in the hydration buffer inlet. (C) FITC-dextran loading efficiencies
corresponding to different FITC-dextran concentrations in the hydration buffer inlet. (D) Mean liposome
radii corresponding to different FITC-dextran concentrations in the hydration buffer inlet.
23
HydrophobicDrugLoading. Nile red loading efficiency increased linearly as FRR increased (Figure
6(A)). Liposome size decreased with increasing FRR (Figure 6(B)). This phenomena can be explained by the
disassembly-reassembly process. During initial liposome formation at the alcohol-buffer interface, Nile red
was trapped into the lipid bilayer due to rapid self-assembly. As the surrounding alcohol concentration
increased above the critical concentration due to diffusion, liposomes partially disassembled into sections
of lipid bilayers with Nile red trapped between the bilayer. Once the surrounding alcohol concentration fell
below the critical concentration due to the mixture of the streams, the sections of lipid bilayers reassembled
together with the initially entrapped Nile red. The higher the FRR, the more rapid the reassembly, and
thereby the smaller liposomes formed as explained in the previous section. The amount of Nile red trapped
in the liposomes depended on the initial liposome self-assembly at the alcohol-buffer interface. At a lower
FRR, initial liposome self-assembly was less rapid and the amount of Nile red trapped within the bilayers
was reduced. The resulting Nile red loading efficiencies were relatively low compared to FITC-dextran
loading efficiencies. This was because Nile red was trapped within the lipid bilayers and thereby took
up only a little space of each liposome. The calculated loading concentrations were therefore significantly
lower than the FITC-dextran loading concentration and resulted in lower loading efficiencies. At a constant
FRR of 30, no correlations were found between TFR and liposome size or loading efficiency (Figure 6(C)(D)).
Liposomes loaded with Nile red were larger than the unloaded liposomes (Figure 6(B)(D)). One possible
reason for the increased liposome size was the less rapid liposome reassembly of the lipid bilayers. Because
of the entrapped Nile red, liposomes disassembled into sections of lipid bilayers instead of lipid molecules.
Since the critical concentration for lipid bilayer sections to self-assemble into liposomes was lower than
that of lipid molecules, the induced self-assembly was less violent and rapid.
24
(A) (B)
(C) (D)
Figure6: Nile red loading. (A) Nile red loading efficiency versus FRR. The lipid mixture inlet flow rate was
controlled at 3µL/min. (B) Mean liposome radius versus FRR when Nile red was loaded. The lipid mixture
inlet flow rate was controlled at 3 µL/min. (C) Nile red loading efficiencies versus TFR at a constant FRR of
30. (D) Mean liposome radius versus TFR at a constant FRR of 30 when Nile red was loaded. The error bar
for each Nile red loading efficiency showed the standard deviation obtained by at least three repeated trials
of the experiment. The error bars for liposome radii showed the polydispersity measured by the Dynamic
Light Scattering.
Concurrent Loading of Hydrophilic & Hydrophobic Drugs. FITC-dextran and Nile red loading
efficiencies did not vary when both drug simulants were loaded concurrently (Figure 7(A)(B)). The fact
that FITC-dextran and Nile red loading efficiencies were independent of one another strengthens the sim-
plicity of tunable loading efficiency by the MHF method. The resulting liposomes were the same sizes as
25
those loaded with Nile red, but larger than those loaded with FITC-dextran liposomes and unloaded lipo-
somes (Figure 7(C)). The reason for the increased sizes of liposomes loaded with Nile red was as discussed
previously.
(A) (B)
(C)
Figure 7: Concurrent loading of 4kDa FITC-dextran and Nile red on liposomes. FRR was varied from
10 to 50 while maintaining a constant lipid mixture inlet flow rate of 3 µL/min. (A) FITC-dextran loading
efficiency when loaded into liposomes alone versus concurrent loading of both drug simulants. (B) Nile red
loading efficiency when loaded into liposomes alone versus concurrent loading of both drug simulants. (C)
Mean liposome radii under different loading conditions. The error bar for each loading efficiency showed
the standard deviation obtained by at least three repeated trials of the experiment. The error bars for
liposome radii showed the polydispersity measured by the Dynamic Light Scattering.
26
2.5 Conclusions
Liposome fabrication and concurrent loading of hydrophilic and hydrophobic drug were demonstrated
using the MHF technique. The effects of FRR on liposome size and loading efficiency were presented.
The results illustrated tunable drug loading efficiency and liposome size by adjusting the input flow rates
of the hydration buffer and lipid mixture. The higher the IPA concentration in the hydration buffer, the
larger the liposomes produced and the lower the hydrophilic drug loading efficiency. The concentration
of the hydrophilic drug did not affect the loading efficiency. The concurrent loading of hydrophilic and
hydrophobic drug simulant was demonstrated to have no impact on the loading efficiency of each drug
simulants. While the loading of the hydrophilic drug simulant showed no influence on liposome size, the
loading of hydrophobic drug simulants resulted in slightly larger liposomes. More studies are required to
support and verify the hypothesized mechanism of liposome formation by the MHF technique. Experimen-
tal evidence indicated that the MHF technique facilitates liposome generation at reduced drug and solvent
consumption, one-step liposome preparation, continuous production, and the potential of scale-up pro-
duction by device parallelization. The ease of generating drug-loaded liposomes using the MHF technique
may obviate the shelf life limitations of liposome preparation to allow implementation in point-of-care
and on-demand liposome fabrication. Future studies in liposome generation at higher lipid concentrations
using the MHF approach and a side-by-side comparison to conventional batch techniques are anticipated.
Supported by the experimental evidence, the MHF technique is promising for industrial-scale drug-loaded
liposome preparation.
27
Chapter3
3D-PrintedMicrofluidicDeviceforHigh-ThroughputProductionof
LipidNanoparticlesIncorporatingSARS-CoV-2SpikeProteinmRNA
Note: This chapter is currently under review for publication as a journal article in ACS Nano (2022).
3.1 Abstract
Lipid nanoparticles (LNPs) are drug carriers for protecting nucleic acids for cellular delivery. The first
mRNA vaccines authorized by the United States Food and Drug Administration are the mRNA-1273 (Mod-
erna) and BNT162b (BioNTech/Pfizer) vaccines against coronavirus disease 2019 (COVID-19). We designed
a 3D printed Omnidirectional Sheath-flow Enabled Microfluidics (OSEM) Device for producing mRNA-
loaded LNPs that closely resemble the Moderna vaccine: we used the same lipid formulations to encapsu-
late mRNA encoding SARS-CoV-2 spike protein. The OSEM device is made of durable methacrylate-based
materials that can support flow rates in the mL/min range and was fabricated by stereolithography (SLA),
incorporating readily adaptable interfaces using commercial fluidic connectors. Two key features of the
OSEM device are: 1) a 4-way hydrodynamic flow focusing region and 2) a staggered herringbone mixer
(SHM). Superior to conventional planar fluid junctions, the 4-way sheath flow channel generates an evenly
focused, circular center flow that facilitates the formation of LNPs with low polydispersity. Downstream,
fluid mixing in the SHM is intensified by incorporating a zig-zag fluidic pathway to deliver high mRNA
28
encapsulation efficiency. We characterized the mRNA-loaded LNPs produced in the OSEM device and
showed that the enhanced 3D microfluidic structures enable a 5-fold higher throughput production rate
(60 mL/min) of LNPs than commercial multi-thousand-dollar micromixers. The device produced LNPs of
diameter less than 90 nm, with low polydispersity (2-8%) and high mRNA encapsulation efficiency ( > 90%).
At a significantly lower cost (US $1.5) compared to commercial instruments, the OSEM device provides an
unprecedented all-in-one solution to LNP production from lab to market.
3.2 Motivation
Lipid-based nanodrug delivery systems play a vital role in the enabling nucleic acid-based vaccines. Nucleic
acid-based vaccines, such as DNA and RNA vaccines, have emerged in recent years particularly due to their
rapid development process relative to other vaccine technologies[59]. Efficient cellular delivery of nucleic
acids relies on cationic lipid-based delivery vehicles that protect them from nuclease degradation[157]. The
first mRNA vaccines authorized by the United States Food and Drug Administration (FDA) are mRNA-1273
(Moderna) and BNT162b (BioNTech/Pfizer) against coronavirus disease 2019 (COVID-19)[61]. Both vac-
cines employ lipid nanoparticles (LNPs) to deliver mRNA encoding the full SARS-CoV-2 spike protein,
providing 95% efficacy in protection against the alpha variant of SARS-CoV-2[130, 5]. mRNA-loaded LNPs
are different from liposomes, which are 30-200 nm spheres made of lipid bilayer(s) surrounding an aque-
ous lumen[96, 157]. LNPs on the other hand have complex internal architectures and are stabilized by
synthetic ionizable cationic lipids and anionic nucleic acids[67]. Recently, Eygeris and colleagues exam-
ined the structure of the LNPs using electron microscopy, resolving multilamellarity and lipid partitioning
[40]. Similar to other nanodrug delivery systems, efficient cellular delivery of lipid-based nanoparticles
requires the particles to be within 30 to 200 nm[38, 179, 147, 31]. This is because particles smaller than
30 nm can leak into liver sinusoidal capillaries[116, 38] and particles larger than 200 nm are susceptible to
29
mononuclear phagocytic uptake[38, 102]. In response to the stringent particle size requirements, a number
of nanoparticle production technologies have been developed for generating size-specific nanodrugs.
Common methods for producing nanoparticles employ either a “top-down" approach or a “bottom-up"
approach[180]. An example of top down approaches for lipid-based nanoparticle formation is sonication,
in which a suspension of lipids in an aqueous phase is disrupted by sonication, breaking lipid floccules
into fragments that rearrange into nanoscale particles via the hydrophobic effect[111]. In bottom up ap-
proaches, the formation of nanoparticles is facilitated by rapidly mixing lipids in a water-miscible solvent
with an aqueous phase[180, 96]. Compared to top down approaches, bottom up approaches enable more
precise control of the particle size [180]. Current mRNA vaccine production employs a microfluidic mixer
that consists of a T-junction followed by a staggered herringbone mixer (SHM) to generate LNPs in a
bottom-up manner[59, 30]. The T-junction first combines a lipid-ethanol stream with an oligonucleotide-
containing aqueous stream, with the SHM inducing rapid convective mixing of the two liquid phases to
facilitate the formation of small LNPs (size ranging from 50 to 140 nm[59]). Compared to macroscopic
mixing methods that are susceptible to high batch-to-batch variations and often produce LNPs with high
polydispersity, microfluidic mixers enable rapid, controlled, and localized convective mixing within a con-
fined micro-environment[13].
Commercial benchtop micromixer instruments cost>US$40,000 [128]. Due to these steep prices, mi-
crofluidic devices that are used in small-scale, preclinical research are mostly fabricated by elastomer mold-
ing. Elaster molding is a laborious microfabrication technique that is limited to small-scale handmade man-
ufacturing. The resulting microfluidic devices often have poor user interfaces, are prone to clog, and are
made of delicate materials that can only be used at low flow rates (around 100 µL/min)[16]. In light of these
challenges, additive manufacturing (i.e. 3D printing) has emerged for microfluidic device fabrication[39, 95,
16, 14]. Additive manufacturing for microfluidic devices offers: 1) low fabrication cost, 2) rapid single-step
manufacturing process from computer designs, 3) transition-free scalability from small-batch prototyping
30
to production, 4) the ability to produce three-dimensional geometries with compact and complex designs,
and 5) the capability to produce readily adaptable interfaces with commercial fluidic connectors[39, 95, 4,
90, 16, 14]. In this work, we designed and printed an Omnidirectional Sheath-flow Enabled Microfluidics
(OSEM) Device that is capable of producing mRNA-encapsulating LNPs at 5-fold higher production flow
rates (60 mL/min) than commercial benchtop micromixers used to produce mRNA vaccines. The OSEM
device produces LNPs in a continuous flow, so the throughput can be further scaled up for large-scale pro-
duction by device parallelization[167]. The materials cost to fabricate one device is US$1.5/device, making
it suitable for both small, research-scale LNP production as well as large, industrial-scale pharmaceutical
production.
Our OSEM device incorporates larger microchannels (500 µm height and width) than those used in
traditional microfluidics ( <100 µm) for increased manufacturability, to prevent clogging, and to increase
throughput. The device is made of durable methacrylate-based photopolymers that support high flow rates
(> 1 mL/min) for high throughput LNP production. In conventional microfluidic devices, successful gen-
eration of pharmaceutic-relevant sized LNPs (30 - 200 nm) relies on the small channel sizes (≤ 100µm)[102,
13, 91]. To generate high quality LNPs with larger microchannels, we implemented advanced 3D struc-
tures to achieve rapid, localized convective mixing. The device channel has two inlets, one for lipid/ethanol
mixtures and another for mRNA in aqueous buffer. The two inflows intersect at a 4-way sheath flow gener-
ator, and then enter a downstream SHM. Based on the LNP formation mechanism proposed by Maeki and
coworkers, intermediate disk-like lipid structures begin to form when ethanol is at a critical concentration
of approximately 60-80%; the amount of time that the lipids are assembled in this intermediate disk-like
structure is key to determining LNP size[102]. The uniformity of the ethanol-water interface along which
the intermediate disk-like lipid structures are formed is therefore important for forming LNPs with low
polydispersity. Traditional 2D microfluidic sheath flow generators split the outer flow into two streams
which then intersect the inner flow stream to focus it in two dimensions. Because the outer flow only
31
intersects with the inner flow from two directions, two walls of the channel are wetted with the inner
stream. The boundary effect of the inner stream near the wall causes heterogeneity in the liquid-liquid in-
terface which is especially pronounced high flow rates, resulting in particles with high polydispersity[66,
65]. This requirement for low flow rates limits the total throughput of the device. Taking advantage of
the capabilities of 3D printing, we have designed a sheath flow generator that splits the outer flow into
four streams which are then intersected with the inner stream at orthogonal planes to focus the flow in
three dimensions. The goal of our upstream 4-way sheath flow is to effectively focus the lipid-ethanol
inflow at the center of the channel to create a uniform ethanol-water interface. This level of uniformity of
the liquid-liquid interface would previously require a delicate concentric glass capillary assembly[66]. As
liquids enter the SHM, the two phases are rapidly agitated by localized convective mixing. In this region,
we expect the intermediate disk-like lipid structures to condense mRNA in the presence of the ionizable
cationic lipids during the assembly of the LNPs.
In this study, we confirmed the OSEM device design by testing devices comprised of different channel
features to produce LNPs. We produced LNPs that resembled the Moderna mRNA-1273 vaccine against
COVID-19 by using the Moderna lipid formulation to encapsulate mRNA encoding for SARS-CoV-2 spike
protein. We demonstrated LNP production at low, research-scale flow rates ( 1− 6 mL/min) as well as high,
production-scale flow rates ( 24− 60 mL/min). The LNPs were characterized by size, zeta potential, and
encapsulation efficiency and were imaged with cryogenic transmission electron microscopy (Cryo-TEM).
Through 3D printed microfluidics, our goal is to provide a low-cost and high-throughput solution that
meets the needs of both small and large scale pharmaceutical LNP production.
32
3.3 MaterialsandMethods
3.3.1 PreparationofLipidMixturesandmRNAEncodingSARS-CoV-2SpikeProtein
Lipid mixture preparation. Lipid mixtures in ethanol were prepared as previously described for the
mRNA-1273 vaccine[140]. Briefly, heptadecan-9-yl-8-((2-hydroxyethyl)(6-oxo-6-(undecyloxy)hexyl)amino)octanoate
(SM-102) (Ambeed Inc, United States of America), 1,2-dimyristoyl-rac-glycero-3-methoxypolyethylene glycol-
2000 (DMG-PEG2000) (Avanti Polar Lipids, United States of America), 1,2-distearoyl-sn-glycero-3-phosphocholine
(DSPC) (Avanti Polar Lipids, United States of America), and cholesterol (Avanti Polar Lipids, United States
of America) were dissolved in ethanol at 50, 1.5, 10, 38.5 mol%, respectively.
PreparationofmRNAencodingSARS-CoV-2spikeprotein. To prepare mRNA molecules resem-
bling those in mRNA vaccines against SARS-CoV-2, a plasmid insert was designed based on the sequence
confirmed by comparative analyses of RNA extracted from original vials of Moderna mRNA-1273 [79]. The
plasmid insert sequence was designed with a BbsI restriction site for plasmid linearization prior to PCR
amplification and a T7 promoter sequence to enable subsequent in vitro transcription (complete insert se-
quence shown in Figure 8). The plasmids (Twist Bioscience, United States of America) were linearized
using BbsI restriction enzymes (New England Biolabs, United States of America) and then PCR amplified
using 3’ and 5’ primers (Integrated DNA Technologies, United States of America) as shown in Figure 8. The
PCR product was then transcribed using a MEGAscript T7 Transcription Kit (Invitrogen, United States of
America) and the mRNA was purified using a MEGAclear Transcription Clean-up Kit (Invitrogen, United
States of America). The sizes of the PCR product and transcription product were confirmed by agarose
gel electrophoresis (Figure 9) using loading dyes and ladders obtained from New England Biolabs, United
States of America. The resulting mRNA was expected to have 3996 bases. mRNA was stored at− 20°C in
25 mM sodium acetate (pH 5) until use.
33
Figure8: Plasmid insert sequence encoding SARS-CoV-2 spike protein and PCR primer sequences. Yellow:
T7 promoter, blue: 3’ untranslated region, hunter green: start codon, brown: signal peptide, orange: spike
protein, red: stop codon, magenta: 3’ untranslated region, light green: BbsI restriction site.
34
(A) (B)
Figure 9: Gel electrophoresis results for (A) PCR and (B) transcription products from pDNA encoding
SARS-CoV-2 spike protein.
3.3.2 DeviceDesignandFabrication
All microfluidic devices were designed using Autodesk Fusion 360 and fabricated using a SLA-DLP printer
(Asiga MAX 27UV, Australia) with Printodent GR-10 resin (Pro3dure Medical, United States of America).
As shown in Figure 10, the 3D printed OSEM device had two inlets: one for lipids and another for mRNA,
and one outlet for collection of the LNP product. The devices were 3D printed at a 10µm layer thickness.
The print duration for 4 devices was 12 hours, averaging a 3 hours/device print time. The resulting mi-
crochannels were washed with isopropanol at 1 mL/min for at least 30 minutes to remove residual uncured
resin within the channels. The device exterior was smoothed to increase device transparency by applying
a thin layer of resin between each surface of the device and a piece of cover glass, postcuring the device
under a 36 Watt UV chamber (ProtoProducts, United States of America) for 30 minutes, and then removing
the cover glass.
35
(A) (B) (C)
Figure 10: (A) Isometric view, (B) front view, and (C) side view of the OSEM device featuring a 4-way
sheath flow channel followed by a downstream SHM for LNP production.
The OSEM device was designed with a 4-way sheath flow channel followed by a SHM (Figure 10(A),
10(B)), all with 500µm sized square-profile channels (detailed design parameters are shown in Figure 11).
Figure 12 demonstrates the hydrodynamic flow focusing effect of the microfluidic channel using water in
both inner and outer flows with red food dye added to the inner flow. At a flow rate ratio of 20 (5 mL/min
clear water vs 0.25 mL/min colored water), the 4-way sheath flow focuses the colored stream at the center
of the channel.
36
Figure11: The OSEM design in drawings showing detailed design parameters. Unit: mm.
Figure 12: Photograph of a 4-way sheath flow demonstrated using water with (inlet flow rate = 0.25
mL/min) and without (inlet flow rate = 5 mL/min) red food dye (FRR = 20).
3.3.3 ProductionofLipidNanoparticles
Microfluidic production of LNPs. Syringes containing the lipid mixtures in ethanol and mRNA in 25
mM sodium acetate (pH 5) were connected to the inlets of the OSEM device as annotated in Figure 10(C)
37
using standard fluidic connectors (IDEX Health & Science, United States of America). Each inlet flow rate
was controlled using a syringe pump (Genie Touch, Kent Scientific, United States of America). Flow rate
ratio (FRR) was defined as the volumetric inlet flow rate of mRNA over the volumentric inlet flow rate of
lipids in ethanol (equation 3.1).
Flow Rate Ratio (FRR) =
Inlet Flow Rate of mRNA Solution
Inlet Flow Rate of Lipid Solution
(3.1)
The mRNA solution was infused at concentrations ranging from 5 to 40 ng/µL and flow rates ranging from
2 to 60 mL/min. Lipid solution was infused at flow rates ranging from 0.1 mL/min to 0.8 mL/min. LNPs
were collected from the outlet of the OSEM device. Total lipid concentrations were adjusted according to
the two inlet flow rates and mRNA inlet concentration to maintain a constant ratio of ionizable polymer
amine (N = nitrogen) groups to negatively-charged nucleic acid (P = phosphate) groups. This ratio (N/P)
was 6, which is identical to that of the Moderna mRNA-1273 vaccine against COVID-19[140].
BufferexchangeandsterilizationoftheLNPs. LNPs collected from the device outlet were buffer
exchanged to 20 mM Tris-Cl buffer (pH 7.5) using 50 kDa MWCO Amicon ™ Ultra-0.5 Centrifugal Filter
Units (MilliporeSigma™, United States of America) and then sterilized using 0.2 µm syringe filters (Pall
Life Sciences, United States of America). Buffer-exchanged LNPs were stored at 4 °C.
3.3.4 LNPCharacterization
Size, Zeta Potential, and Particle Concentration. LNP size, polydispersity index (PDI), and zeta po-
tential were obtained from dynamic light scattering (DLS) (Malvern ZetaSizer Ultra, Malvern Panalytical,
United Kingdom). Each data collection run was repeated at least 3 times; standard deviations across runs
are reported in the Results & Discussion section below. Nanoparticle tracking analysis (NanoSight NS300,
Malvern ZetaSizer Ultra, Malvern Panalytical, United Kingdom) was used for particle concentration mea-
surments.
38
mRNAEncapsulationEfficiency. The mRNA encapsulation efficiency of the LNPs is defined as the
fraction of mRNA encapsulated within the particles over the total mRNA in solution. This was calculated
by measuring the fraction of unencapsulated mRNA over the total mRNA in a sample of LNPs (equation
3.2). Unencapsulated mRNA is free in solution while total mRNA can be freed by lysing the LNPs. The
amount of mRNA was determined using RiboGreen fluorescence assay (Quant-it ™ RiboGreen RNA Assay
Kit, Invitrogen, United States of America) according to the manufacturer’s protocol. The total mRNA in
solution was determined by lysing the LNPs in the presence of 1% Triton™ X-100 (Thermo Scientific, Unites
States of America) for 10 minutes at 37°C prior to binding. Fluorescence measurement was performed using
BioTek Synergy H1 Microplate Reader (Agilent, United States of America). Each data point was obtained
in duplicate.
% Encapsulation Efficiency = 1− unencapsulated mRNA
total mRNA
× 100% (3.2)
Cryo-TEM Imaging. Based on the optimized parameters obtained as described below, LNPs were
produced at FRR = 20, mRNA inlet concentration = 10 ng/µL, and mRNA inlet flow rate = 5 mL/min, and
then concentrated 50 times using 50 kDa MWCO Amicon™ Ultra-0.5 Centrifugal Filter Units (Millipore-
Sigma™, United States of America). 8% sucrose was added to the samples as a cryoprotectant. In stained
samples, 1% uranyl acetate (pH 4.0) (Ted Pella, Inc. United States of America) in water was mixed with
the sample immediately before freezing. The sample was applied to glow-discharged C-flat ™ holey carbon
grids (CF-2/1-2Cu-50, Electron Microscopy Sciences, United States of America) at 22°C and 100% relative
humidity using an FEI Vitrobot Mark IV (ThermoScientific, United States of America). We first applied a
4µL drop of sample to the grid, waited 10 minutes, and gently blotted the grid. Then, we applied another
4 µL drop of sample, waited 15 seconds, blotted the grid, and vitrified the sample into liquid ethane. The
temperature of the sample was kept below -170°C at all times. The samples were imaged using a Talos™
F200C TEM (ThermoScientific, United States of America) operating at 200 kV. Images were acquired with
a 4k× 4K Ceta CMOS camera at 36,000x magnification.
39
3.3.5 COMSOLMultiphysicsFlowSimulation
Ethanol concentration profiles in hydrodynamic flow focusing channels were simulated using COMSOL
Multiphysics (version 5.6) with 3D channel geometries that are identical to the sheath flow region of the
3D-printed microfluidic devices (both 2-way and 4-way sheath configurations). The ethanol diffusivity in
water was set toD = 1.6× 10
− 10
m
2
/s[62].
3.4 Results&Discussion
ProductionofmRNA-loadedLNPsusingtheOSEMdevicefeaturinga4-wayhydrodynamicflow
focusing channel followed by a downstream SHM. To evaluate the impact of the 4-way sheath flow
and the SHM on mRNA-loaded LNPs, we produced LNPs using three devices, each implemented with
different combinations of key microfluidic features: 1) the OSEM device which is a 4-way sheath flow device
with downstream SHM (4-Way Sheath Flow, Figure 10), 2) a 2-way sheath flow device with downstream
SHM (2-Way Sheath Flow, Figure 13(A)) and 3) a a 4-way sheath flow device without downstream SHM
(No Mixer, Figure 13(B)). DLS results show that larger LNPs were produced using the “2-way sheath flow
device" (diameter = 119.67± 1.61 nm) and the “no mixer device" (diameter = 124.50± 0.436 nm) than
the “4-way sheath flow device" (diameter = 103.77 ± 4.41 nm), and that the “2-way sheath flow device"
produced LNPs with significantly higher polydispersity (PDI = 0.131 ± 0.001) compared to LNP produced
with the other two devices (PDI = 0.0478± 0.0202 and 0.0568± 0.0168 for the “4-way" and “no mixer"
devices respectively, Figure 14(A)). Among the three devices, the “4-way sheath flow device" produced
LNPs with the highest encapsulation efficiency of 88.56%, whereas the “no mixer device" produced LNPs
with the lowest encapsulation efficiency of 61.58% (14(B)). These results suggest that the 4-way sheath
flow feature is critical in producing LNPs with low polydispersity and that the SHM facilitates efficient
encapsulation of mRNA in LNPs.
40
(A) (B)
Figure 13: Photographs of 3D-printed microfluidic devices featuring (A) a 2-way sheath flow channel
followed by a downstream SHM and(B) a 4-way sheath flow channel without downstream mixers.
(A) (B)
Figure14: (A) Size and(B) mRNA encapsulation efficiency of LNPs produced using a 4-way sheath flow
channel with downstream SHM device (4-Way Sheath), a 2-way sheath flow channel with downstream
SHM device (2-Way Sheath), and a 4-way sheath flow channel without the downstream SHM device (No
Mixer). LNPs were produced at a total inlet mRNA flow rate of 4 mL/min, mRNA inlet concentration of 5
ng/µL, and FRR of 20. Error bars show standard deviations across> 3 repeated trials. Photographs of the
2-Way Sheath Flow device and the No Mixer device are shown in Figure 13.
To further understand the impact of the 4-way sheath flow on LNP production, we simulated the hy-
drodynamic flow focusing of ethanol in water in devices featuring 4-way and 2-way sheath flow channels
using COMSOL Multiphysics (Figure 15). Both simulations were performed assuming an inlet ethanol flow
rate of 0.25 mL/min and a total inlet water flow rate of 5 mL/min (FRR = 20). As shown in Figure 15(H),
41
the 4-way sheath flow resulted in a focused circular ethanol stream at the center of the channel in the x-z
plane. On the other hand, the 2-way sheath flow channel did not result in a focused ethanol stream (Figure
15(D)); ethanol was concentrated at the corners of the channel instead of being focused at the channel
center as observed at a low flow rate (total water inlet flow rate = 5 µL/min, Figure 16). In the 4-way
sheath flow channel, a uniform, circular ethanol and water liquid-to-liquid interface was created, enabling
the production of nanoparticles with low polydispersity at high flow rates.
42
Figure15: COMSOL Multiphysics simulation results showing ethanol concentration in water and outlet
flow velocity within hydrodynamic focusing channels. The FRR was set to 20 and the total water inlet
flow rate was set to 5 mL/min. (A) &(E) Images of the 2-way and 4-way sheath flow region of the device,
respectively. Ethanol concentration results for a 2-way sheath flow channel in (B) the x-y plane at the
center of the channel, (C) an orthogonal view of showing concentration profiles in x-z planes, and (D)
the x-z plane at outlet y = − 3 mm with flow velocity contours. Ethanol concentration results for a 4-
way sheath flow channel in (F) the x-y plane at the center of the channel, (G) orthogonal view showing
concentration profiles in x-z planes, and(H) thex-z plane at outlety =− 3 mm with flow velocity contours.
43
Figure16: COMSOL Multiphysics simulation results showing ethanol/water hydrodynamic focusing at 5
µ L/min total water inlet flow rate. The FRR was set to 20. (A) &(D) Results of a 2-way and a 4-way sheath
flow regions of the device respectively, in the x-y plane. (B) &(E) Results for a 2-way and a 4-way sheath
flow channel, respectively, in orthogonal view showing slices of the concentration profile in x-z planes.
(C) & (E) Results for a 2-way and a 4-way sheath flow channel, respectively, in the x-z plane at outlet
y =− 3 mm with flow velocity contours.
These results align with LNP formation mechanism described by Maeki and coworkers[102]. In the
hydrodynamic flow focusing region, lipids form intermediate disk-like structures along the ethanol-water
44
interface at the critical ethanol concentration[102]. The formation of these intermediate disk-like struc-
tures determines particle size[102]. The 4-way sheath flow created a more uniform, circular ethanol-water
interface than the 2-way sheath flow, which enabled a more even and controlled diffusion from the lipid-
ethanol stream to the aqueous stream (Figure 15). Therefore, devices that implemented 4-way sheath flow
produced LNPs with lower particle polydispersity compared to those produced with with 2-way sheath
flow (Figure 14(A)). In the SHM, the intermediate disk-like cationic lipid structures condensed with polyan-
ionic mRNA and formed LNPs under rapid, localized, convective mixing. Due to rapid mixing, LNP for-
mation was completed in a short distance. The homogeneity of the mixture at time of LNP formation
therefore impacted the mRNA encapsulation efficiency of the LNPs. This explains why the “no mixer" de-
vice produced LNPs at significantly lower encapsulation efficiency that devices implementing SHM. Our
experimental and simulation results together suggest that the superior focusing of the 4-way sheath flow
over 2-way sheath flow enabled the formation of LNPs with lower polydispersity, and that the downstream
SHM is critical to achieve high mRNA encapsulation efficiency.
ImpactofproductionparametersoftheOSEMdeviceonLNPcharacteristics:size,zetapoten-
tial, and encapsulation efficiency. To understand how each production parameter impacts the LNPs,
we used the OSEM device, which features optimized device configuration (4-way sheath flow, SHM), to
produce LNPs at varying inlet concentrations, inlet flow rates, and flow rate ratios (Figure 17). The results
show that LNP size increased as inlet concentration increased (Figure 17(A)). A typical mRNA vaccine
preparation pipeline includes an additional particle concentration step using centrifugal filters to adjust
the final vaccine dosage[59]. These centrifugal filters have high concentration factors ( e.g., 80–100X, Am-
icon Ultra 15 mL Filters, EMD Millipore, Billerica, MA, USA). Therefore, our demonstration of loading 40
µg/µL of mRNA onto LNPs using the OSEM device suggests that, with a downstream particle concentra-
tion step, the device is highly applicable for producing vaccines with a final mRNA dosage at 100 µg per
0.5 mL (500 µg/mL), i.e. the dosage administered in the Moderna vaccine. LNP size decreased as the total
45
flow rate was increased from 1 mL/min to 6 mL/min (Figure 17(B)). This result aligns with the formation
mechanism proposed by Maeki and coworkers[102], in which the duration of the slow, diffusion-driven
mixing significantly impacts the LNP size due to the formation of intermediate disk-like lipid structures. At
higher flow rates, the duration of the diffusion-driven mixing is shorter, reducing the formation of disk-like
lipid structures and therefore resulting in smaller particles. The shortened diffusion-driven mixing period
prevents the formation of disk-like lipid structures, causing the LNPs to be assembled further downstream,
where the mRNA is more homogeneously mixed. Therefore, higher mRNA encapsulation efficiencies were
achieved as the mRNA is condensed in the presence of the synthetic ionizable cationic lipids (Figure 17(B)).
Additionally, we demonstrated production of LNPs at high throughput by increasing the mRNA inlet flow
rate from 24 mL/min to 60 mL/min. Within this range of flow rates, the total flow rate did not significantly
impact the LNP size. Produced at an mRNA inlet flow rate of 60 mL/min, the LNPs had an average diameter
of85.68± 0.81 nm, polydispersity of6.42%± 0.59%, zeta potential of− 3.78± 1.00 mV, and mRNA en-
capsulation efficiency of 93.28%± 8.28% (Figure 17(B)). This suggests that the OSEM device is capable of
producing mRNA vaccines at a 5-fold higher throughput compared to commercial, multi-thousand-dollar
micromixers[162]. The volumetric throughput can be further increased by assembling multiple devices in
parallel.
In the FRR-varying experiment, the combined mRNA and lipid inlet flow rate was fixed at a constant
4.2 mL/min. The FRR did not have significant impact on LNP size (Figure 17 (C)). The consistent LNP
sizes resulting from varying FRRs can be explained by the successful centering of the lipid stream in the
4-way sheath flow channel. As Maeki and coworkers described, LNP size is determined during the slow,
diffusion-driven upstream mixing[102]. Within the range of FRRs we worked with, the lipid stream is
focused in the channel center, creating a narrow concentration gradient following the channel intersection.
Therefore, consistent LNP sizes were observed. This differs from previous observation in conventional 2-
way hydrodynamic flow focusing studies[96, 102] due to the boundary effects in planar flow focusing
46
channels (Figure 15(D)). As FRRs in planar devices impact the surface contact of the center stream with
the channel wall, the resulting particle sizes varied accordingly[102]. Our results show that the LNPs
can be produced at high FRRs (up to 30) at which the final ethanol content of the LNP is low. This is
favorable for the subsequent steps of vaccine production, during which ethanol needs to be removed via
buffer-exchange techniques such as centrifugal filtration or dialysis[59]. All zeta potential data was near
neutral due to the presence of PEGylated lipid, which created an electrostatic barrier at the surface of the
particle. All sets of parameters tested produced LNPs with encapsulation efficiencies > 83% (Figure 17),
polydispersity< 8%, and average diameters ranging from 85 - 130 nm (Figure 17). We showed that LNPs
can be produced at flow rates ranging from 1 mL/min to 60 mL/min (Figure 17(C)). Compared to current
commercial microfluidic mixers used for mRNA vaccine production, which produced 50-140 nm mRNA-
loaded LNPs with>69% encapsulation efficiency with < 12 mL/min production flow rate[59, 162], our 3D
printed OSEM device offers a 5-fold higher throughput with superior performance at a significantly lower
cost (US$1.5 materials cost per device).
47
Figure17: DLS size (column 1), zeta potential (column 2), and encapsulation efficiency (column 3) results
for LNPs produced under various conditions. (A) Results for LNPs produced at mRNA inlet concentration
ranging from 5 to 40 ng/µL, at constant FRR of 20 and inlet mRNA flow rate of 4 mL/min. (B) Results for
LNPs produced at inlet mRNA flow rate ranging between 2− 6 mL/min (small-scale) and24− 60 mL/min
(high throughput), at constant FRR of 20 and inlet mRNA concentration of 10 ng/µL. LNPs produced at
5 mL/min mRNA inlet flow rate (highlighted) were imaged using cryo-TEM as shown in Figure 18. (C)
Results for LNPs produced at FRR ranging from 5 to 30, at constant total flow rate of 4.2 mL/min, inlet
mRNA concentration of 10 ng/µL. Error bars of standard deviations among> 3 repeated trials were plotted.
We selected the LNPs produced at mRNA inlet flow rate of 5 mL/min, mRNA inlet concentration of
10 ng/µL, and FRR of 20 for Cryo-TEM imaging (Figure 18). These LNPs had an average diameter of 89.76
± 2.4 nm, PDI of 0.0195± 0.0152, zeta potential of− 3.41± 0.83 mV, and encapsulation efficiency of
90.3± 4.4 % (data highlighted in Figure 17(B)). The LNP size remained stable after 30 days under storage
at 4°C (Figure 19). Aside from DLS, we measured the particle concentration using nanoparticle tracking
analysis. At an inlet mRNA concentration of 10 ng/µL and FRR = 20, LNPs were produced at a (6.91± 48
0.111)× 10
10
particles/mL. The size measurement from nanoparticle tracking analysis was similar to the
DLS data (diameter = 94.9± 1.4 nm (standard error of the mean)). As shown in Figure 18(B), many of
the LNPs displayed multi-compartment structures consistent with previous descriptions of mRNA vaccine
LNPs [91, 140]. The size of LNPs imaged by Cryo-TEM was consistent with DLS and nanoparticle tracking
analysis. Cryo-TEM images show a clearly resolved lipid bilayer bounding the particles; uranyl acetate-
stained particles show the encapsulated mRNA/lipid structure (Figure 18).
(A) (B)
Figure 18: Cryo-TEM images of LNPs produced by the 3D-printed microfluidic device at FRR = 20, inlet
mRNA concentration = 10 ng/µL, and inlet mRNA flow rate = 5 mL/min. (A) LNPs in 20mM Tris-Cl buffer
(pH 7.5). (B) LNPs stained with 1% uranyl acetate (pH 4.0).
49
Figure19: DLS size measurment results of LNPs on the day they were produced vs 30 days after storage
at 4°C. The LNPs were produced at FRR = 20, inlet mRNA flow rate of 5 mL/min, and inlet mRNA concen-
tration of 10 ng/µL. Error bars show standard deviations among> 3 repeated trials.
3.5 Conclusion
We designed and printed an OSEM device to produce mRNA-encapsulating lipid nanoparticles. In a
3D printable format, our microfluidic device is easier to produce than traditional small-batch elastomer-
molded microfluidic devices, which are known to have poor user interfaces[83, 25] and involve labor-
intensive microfabrication processes[110, 16]. To confirm the device performance in producing clinically-
relevant LNPs, we prepared lipid/mRNA particles similar to the those comprising the Moderna mRNA-1273
vaccine against COVID-19 [79] using the same lipid formulations as the vaccine and an mRNA sequence
encoding the SARS-CoV2 spike protein [140]. We characterized the LNPs in terms of size, zeta potential,
and encapsulation efficiency.
Our microfluidic channel has a 4-way sheath flow channel to focus the lipid-ethanol stream at the
center of the channel, followed by a downstream SHM region that rapidly mixes the two liquid streams
(Figure 10&12). We showed that the two key features of the OSEM device, the 4-way sheath flow and
SHM, are essential to the formation of small LNPs with low polydispersity and high mRNA encapsulation
efficiency (Figure 14). The ethanol concentration profile across the hydrodynamic flow focusing region
50
was simulated using COMSOL Multiphysics, confirming the circular focusing of the ethanol stream at the
channel center (Figure 15).
We tested different mRNA-LNP production parameters and showed that particle size can be tuned
by adjusting the total inlet flow rate and inlet concentrations (Figure 17). Cryo-TEM imaging showed
that some of the LNPs have multi-compartment structures as seen in previous studies (Figure 18). The
OSEM device is capable of producing small LNPs (< 90 nm) with low polydispersity (< 7%) and high
encapsulation efficiency ( > 93%) at 60 mL/min for large-scale mRNA vaccine production. At a material
cost of US$1.5/device and an average print time of 3 hours/device, the OSEM device is more cost-effective
than several-thousand-dollar benchtop commercial micromixers (US$40-45k) [128, 162, 59, 6, 118]. In
addition to the device fabrication cost, the infrastructure cost including the commercial fluidic connecters,
tubing, syringes, and two syringe pumps used in this work cost no more than US$1,200. Most of the cost
(approximately US$950) was spent on syringe pumps, which are readily available in common research
laboratories. Considering the infrastructure cost, the total cost to assemble the system for producing LNPs
using the 3D printed OSEM device remains significantly lower than commercial benchtop micromixers,
delivering an unparalleled solution for applications ranging from small-scale, pre-clinical studies to large-
scale pharmaceutical production.
51
Chapter4
KineticOff-RateSelectionsUsinga3D-PrintedMicrofluidicDevice
Note: This chapter has been published as a journal article. The full reference is: William E. Evenson et al.
“Enabling Flow-Based Kinetic Off-Rate Selections Using a Microfluidic Enrichment Device”. In: Analytical
Chemistry 92.15 (Aug. 4, 2020). Publisher: American Chemical Society, pp. 10218–10222. issn: 15206882.
doi: 10/gqjqcq
4.1 Abstract
Modern genomic sequencing efforts are identifying potential diagnostic and therapeutic targets more
rapidly than existing methods can generate the peptide- and protein-based ligands required to study them.
To address this problem, we have developed a microfluidic enrichment device (MFED) enabling kinetic off-
rate selection without the use of exogenous competitor. We tuned the conditions of the device (bed volume,
flow rate, immobilized target) such that modest, readily achievable changes in flow rates favor formation or
dissociation of target-ligand complexes based on affinity. Simple kinetic equations can be used to describe
the behavior of ligand binding in the MFED and the kinetic rate constants observed agree with independent
measurements. We demonstrate the utility of the MFED by showing a 4-fold improvement in enrichment
compared to standard selection. The MFED described here provides a route to simultaneously bias pools
52
toward high-affinity ligands while reducing the demand for target-protein to less than a nanomole per
selection.
4.2 Motivation
Affinity reagent generation is at the core of developing diagnostics and therapeutics. Recent advance-
ments in cancer genetics have helped to rapidly identify potential molecular targets on a genomic scale.
Accelerated reagent generation is necessary to keep pace with the growing rate at which potential targets
are identified. mRNA display is a selection technique that offers a rapid and cost-effective route towards
generating reagents ranging in size from peptides[75, 77, 76] to proteins[113, 171, 55, 122, 121, 92, 24] and
antibodies[36]. In mRNA display, randomized libraries of polypeptides are covalently attached to their en-
coding mRNA, enabling wholly in vitro selection and directed evolution. To do this, a naïve initial library
is subjected to a cyclic process involving selective enrichment and amplification. The result is that rare
functional molecules from the initial library are enriched over several rounds of selection.
Most selection experiments aim to maximize the affinity (K
D
) of the ligand for the target. Diffusion lim-
its the on-rate of neutrally charged peptides to around10
4
− 10
6
M
-1
s
-1
while off-rates tend to vary much
more[139]. Maximizing affinity therefore involves finding complexes with the slowest dissociation kinet-
ics (K
off
) while maintaining near diffusion-limited on-rates. In prior work, Boder and Wittrup devised a
strategy for improving the off-rates of pools using the target as a competitor in binding selections[18]. This
“off-rate selection” has resulted in very high affinity binders, including a femtomolar fluorescein antibody
and single digit picomolar peptide binders to B-cell Lymphoma-extra Large (Bcl-x
L
, a pro-survival protein
often overexpressed in cancer cells.)[76] In this approach, a vast excess of free target is added (typically
100-fold) to create a condition where any ligand that dissociates from the target is lost, thereby biasing
53
the pool toward molecules with lower dissociation rate constants. Competitor-driven screening thus lim-
its optimal ligand development to systems where milligrams of target are readily accessible, significantly
hindering genomic-scale ligand development.
Other techniques have been developed to perform off-rate selections without additional competitor.
These include extensive washing and volume dilution techniques[100, 101, 132, 120]. Extensive washing
removes ligands as they dissociate from beads with manual washes. The volume dilution technique relies
on a large volume of buffer that is added to the beads after an initial binding step. This drastically decreases
the concentration of target and ligand to impair the rebinding of lower affinity ligands as they dissociate
from the target. In this chapter, we introduce a new competitor-free off-rate selection technique that offers
additional kinetic control compared to extensive washing and volume dilution techniques.
We have implemented a flow-based strategy enabling kinetic off-rate selections without exogenous
competitor. To do this, we designed a microfluidic enrichment device (MFED) consisting of 3D printed
parts, a microfluidic channel, and a frit to enable experiments with non-magnetic beads. We tuned the
conditions of the device (bed volume, flow rate, immobilized target) such that modest, readily achievable
changes in flow rates favor formation or dissociation of target-ligand complexes based on affinity. At low
flow rates, a 5 µ L bead bed volume with 100 pmol immobilized target rapidly captures mRNA displayed
peptides similar to manual pull-down experiments because the residence time on the bed is sufficient to
facilitate binding. Increasing the flow 20-fold decreases the residence time such that there is not enough
time for the library to diffuse to the surface of the beads and bind. By varying the duration of bead washing
after binding, we can enrich the library in binders with slower off-rates where the complex stays intact
longer. Thus flow, rather than competitor, creates the conditions needed to perform an off-rate selection.
Simple kinetic equations can be used to describe the behavior of ligand binding in the MFED and the
kinetic rate constants observed agree with independent measurements. We demonstrate the utility of the
MFED by showing a four-fold improvement in enrichment compared to standard selection using previously
54
identified Bcl-x
L
ligands. The MFED described here provides a route to simultaneously bias pools toward
high-affinity ligands while reducing the demand for target-protein to less than a nanomole per selection.
4.3 MaterialsandMethods
4.3.1 MFEDDesignandFabrication
The MFED consisted of a set of 3D printed parts that housed a polyether ether ketone (PEEK, outer ring
material: polychlorotrifluoroethylene, PCTFE) frit with 10 µ m pore size (IDEX Health & Science). The 3D
printed parts of the microfluidic device were fabricated using a benchtop digital light processing stere-
olithography (DLP-SL) printer (Asiga Max X27 UV, Asiga, Australia) with a clear methacrylate-based resin
(Pro3dure GR-1 Clear, Pro3dure Medical, Germany). The MFED connected two segments of microfluidic
polytetrafluorethylene (PTFE) tubing (1/16" outer diameter and 1/32” inner diameter) to both sides of the
frit (Figure 20) to facilitate continuous flow across the frit. In order to minimize interactions with the 3D
printed parts, the MFED only allowed the liquid to come in contact with the tubing and the frit. The frit
was used to immobilize agarose beads (50-150µ m, Thermo) on one side while allowing free ligands to flow
past the bead bed.
(A) (B) (C)
Figure 20: Design of the MFED. (A) Schematic illustration of bead washing on the MFED. (B) Picture of
the assembled MFED. (C) Picture of the disassembled MFED. 1 cm scale bar indicated.
55
4.3.2 LigandPreparation
We used two previously identified Bcl-x
L
ligands: E1 (amino acid sequence MIETITIYNYKKAADHFSMSM)
and Pep2 (MWRWKMIADQL)[75, 76]. E1 was previously characterized to be a high-affinity ligand with
a binding constant (K
d
) of 20-40 pM and an off-rate constant of 7.4× 10
− 6
s
− 1
. Pep2 was previously
characterized to be a lower affinity ligand when compared to E1, with a K
d
of approximately 65 nM. Both
clones were amplified by polymerase chain reaction (PCR) with Taq polymerase using the same primers
(5’ = TAATACGACTCACTATAGGGACAATTACTATTTACAATTACA and 3’ = GCTGGAGCCACTGCCA-
GATCCCA) and then transcribed in vitro using T7 RNA polymerase[99]. Upon completion, transcrip-
tion was stopped by adding 10% (v/v) of 500 mM EDTA (pH=8) and the mixture was buffer exchanged
into Milli-Q water using 30K Amicon filters (EMD Millipore). The mRNA was ligated to F30P (phos-
phate–dA21–[C9]3–dAdCdCP, synthesized at the Keck Oligo Facility at Yale using reagents from Glen
Research) using T4 DNA ligase. The ligated mRNA was purified by urea-PAGE and resuspended in Milli-
Q water. Subsequently, the clones were translated with 400 nM ligated mRNA in 20 mM HEPES-KOH
(pH = 7.6), 100 mM KOAc, 0.5 mM Mg(OAc)
2
, 8 mM creatine phosphate, 2 mM DTT, 25µ M each amino
acid, and 40% rabbit reticulocyte lysate (Green Hectares; prepared according to the method of Jackson and
Hunt[71]). The translation reactions were incubated at 30 ° for 1 hr, followed by fusion formation with
an addition of 250 mM KCl and 30 mM MgCl
2
while incubating at room temperature for 5 minutes. For
the translation of radiolabeled peptides, unlabeled methionine was replaced with
35
S methionine (Perkin
Elmer).
After translation, the fusions were purified using poly T agarose beads. Biotinylated poly T (IDT) was
immobilized on Streptavidin agarose beads (Thermo). 500µ L of 100 mM HEPES-KOH (pH=8), 1 M NaCl,
0.2% (v/v) Triton-X-100 with poly T beads was added to 100µ L of translation and incubated at 4°C for 1 hr
with rotation. Subsequently, the beads were washed on a Spin-X filter (Corning) and eluted with Milli-Q
water. The eluted fusions were reverse transcribed with Superscript IV in 50 mM Tris-HCl (pH=8.3), 75
56
mM KCl, 3 mM MgCl
2
, 10 mM DTT, 500µ M each dNTP, and 1 mM 3’ primer. The reaction was incubated
at 42°C for 1 hour to facilitate reverse transcription, then further incubated at 65°C for 15 min to inactivate
Superscript IV. The resulting reverse-transcribed fusions (RT’d fusions) were kept in the fridge until use
in radioactive binding assays or selection.
4.3.3 TargetProtein
Bcl-x
L
was gifted from Terry Takahashi and Farzad Jalali-Yazdi[76], who obtained the gene for the first
209 amino acids of Bcl-x
L
(Clone HsCD00004711; Dana Farber/Harvard Cancer Center DNA Resource
Core). The gene was PCR amplified with Pfusion polymerase and was added to an N-terminal avitag
(AGGLNDIFEAQKIEWHEGG) forinvivo biotinylation. The product was cloned into pET24a and expressed
at 37°C overnight in BL21(DE3) cells using an auto-induction media[152]. The cells were lysed with Bper
(Pierce) and purified on Ni-NTA resin on an FPLC (Bio-Rad). The fractions were combined, concentrated,
and then reacted with BirA biotin ligase to tag the protein with biotin. The protein was buffer exchanged
into 1X PBS, frozen in liquid nitrogen, and stored at -80°C.
4.3.4 RadioactiveBindingAssays
100 pmol of biotinylated Bcl-x
L
were coupled to 5 µ L of neutravidin agarose beads (Thermo Fisher Sci-
entific, USA) in a binding buffer at 4 °C for 1 hour. The binding buffer consisted of 1X PBS with 0.1%
(w/w) bovine serum albumin (BSA) and 0.01% (w/w) yeast tRNA. Subsequently, the beads were washed
and blocked in blocking buffer (binding buffer + 10 µ M biotin). The MFED was positioned vertically while
the beads were withdrawn into the device using a syringe pump (Next Advance pump, BD syringe) through
the inlet of the device at a flow rate of 500 µ L/min, packing the beads against the frit (Figure 20(A)). Roughly
100,000 counts of radiolabeled reverse transcribed fusions (RT’d fusions) of E1 or Pep2 were then diluted
in a 1:3 ratio in the blocking buffer and withdrawn into the MFED through the inlet at various flow rates
57
to bind the immobilized Bcl-x
L
in the device. The flow-through was collected and the radioactive counts
were recorded. After binding, blocking buffer was infused through the inlet of the device to wash the
beads at various flow rates and each wash was collected at the outlet for radioactive counting. The beads
were removed with 1500µ L/min blocking buffer in the opposite direction. The beads were collected and
counted on a scintillation counter (Beckman LS 6500).
4.3.5 Selection
Selection on the MFED was performed in the same manner as the radioactive binding assays with a few
exceptions: 1) the flow-through and washes were not collected, and 2) the beads were evacuated with a PCR
solution instead of blocking buffer in order to PCR amplify the bound ligands afterwards. In addition to
MFED-based selection, control trials were performed on Spin-X columns. In the control trials, the ligands
were allowed to bind the beads in 500µ L of binding buffer for 1 hr at room temperature, and then washed
3 times on a Spin-X column with binding buffer.
4.3.6 PCR
After selection, the beads resuspended in a PCR reaction solution were amplified using standard thermo-
cycling techniques. After 9 cycles, the PCR products were run on an agarose gel to determine the presence
of visible bands. If no bands were visible, the reaction was run for an additional 3 cycles. If a weak band
was observed, 1 to 2 additional cycles were performed. After additional cycles, the PCR products were
again run on an agarose gel to determine if additional cycles were required to attain visible bands. After
selection, samples were amplified between 12 and 16 cycles in order to observe bands.
58
4.3.7 BandIntensityAnalysis
The E1 PCR product was 126 base pairs (bp) and the Pep2 product was 96 bp. The intensities of each band in
a 2% agarose gel were observed to determine relative abundances of each clone before and after selection.
The band intensity ratios of E1 and Pep2 were obtained using Image Studio Lite and the corresponding
molar ratios were determined using a calibration curve which correlates molar ratios and band intensity
ratios.
4.4 ResultsandDiscussion
TheMFEDisdesignedtoloadandwashbeadsincontinuousflow. The MFED is designed to enable
ligand capture and washing on beads (magnetic or non-magnetic) large enough to be trapped in the frit.
The device tested here consists of a set of 3D printed parts and a 10µ m pore size frit (Figure 20).
We have previously demonstrated the capacity of 3D printing to fabricate complex microfluidic device
morphologies[15]. This device is fabricated using a benchtop digital light processing stereolithography
printer from a clear methacrylate-based resin. The device connects two segments of microfluidic tubing
(1/16” inner diameter) to either side of the frit. Flow is driven by a syringe pump over a wide range of rates
(from > 1µ L/min to < 2 mL/min).
Flowrateandresidencetimedetermineligandbindinganddissociation. The MFED is designed
to function in two modes — to load ligand (here, mRNA display libraries) onto target-modified beads and
to wash nonspecific and weak binders off the bead bed. For loading, the flow rate must be adjusted such
that the second-order binding reaction can reach equilibrium during the time the ligand transits the bed.
To test this process, we measured the binding of a high affinity radiolabeled peptide ligand previously
described (E1 peptide, K
d
= 40 pM) to immobilized Bcl-x
L
[76]. In order to mimic the enrichment that
occurs in a selection, the E1 peptide was constructed in mRNA display format — as an mRNA-peptide
59
fusion where the C-terminus of the peptide is covalently attached to its mRNA via a puromycin bearing
linker. We then explored the formation of the E1 ligand-Bcl-x
L
complex under different flow conditions
(Figure 21). The E1 peptide has an extremely slow dissociation rate constant (k
off
= 7.4× 10
− 6
s
− 1
),
such that >95% of complexes formed should remain stable during the course of any experiment. At low
flow rates (<12.5 µ L/min) E1-Bcl-x
L
complex formation is essentially quantitative. However, as the flow
is increased, we observed a steady decrease in fraction of E1 peptide binding the immobilized target, with
very little binding at flow rates of 500 µ L/min and higher.
Figure 21: Characterizing and modeling E1 peptide-mRNA fusion binding at various flow rates. The
percent bound increases as the flow rate decreases and plateaus at 73%. Error bars represent the standard
deviation of percent bound over three trials. The best-fit model gives k
on
= (2.5± 4)× 10
4
M
− 1
s
− 1
.
The binding vs. flow curve is essentially the mirror image of the E1-Bcl-x
L
formation reaction, with
low flow rates giving a residence time on the bead bed long enough for the reaction to go to completion
and high flow rates providing a short enough residence time for almost no complex formation. The curve
can be fit to a pseudo-first-order on-rate model, under the condition of excess target where ligand flow is
directed towards a bead bed. The residence time (τ ) of ligand flowing through the bead bed is a function
of volume of the bead bed (V), the packing efficiency of the beads ( η ), and the flow rate (Q):
τ =
V(1− η )
Q
(4.1)
60
The bead packing efficiency ( η ), was estimated as 74%, the close packing density of regular spheres, giving
τ values of 6 seconds for 12.5µ L/min and 150 msec for 500µ L/min, respectively.
Since the dissociation of E1 is much slower than the time scale of these experiments, we can model
the percentage of ligand bound (B) as a function of maximum ligand binding (B
max
), association constant
(k
on
), initial target concentration ([T]
0
), and residence time (τ ):
B = B
max

1− e
− k on· [T]
0
·τ
(4.2)
This analysis givesk
on
= (2.5± 0.4)× 10
4
M
− 1
s
− 1
, which is in agreement with prior work on the diffusion
limit of ligands[139]. The B
max
of 75% is typical for peptides constructed from oligos as mRNA display
fusions monitored by radioactive labeling. The fact that B
max
is not 100% is likely due to chemical synthesis
errors in template construction, with 75% efficiency corresponding to an error rate of ∼ 0.5% per position
((0.995)
63
= 73%, insertions, deletions, and mutations).
High flow rates mimic the conditions needed for “optimal” selection of ligands. Boder and
Wittrup used off-rate selections to maximize the binding free energy in directed evolution experiments.[18,
17] The key to their optimization was establishing conditions where ligands (library members) that dis-
sociate from the target cannot rebind. Their work achieved this condition by adding a large molar excess
(100X) of free target as competitor. Here, the ligand binding vs. flow experiments (Figure 21) demonstrate
that the no-rebinding condition can also be met at high flow rates, without using exogenous competitor.
This is an important observation because it can drastically reduce the amount of target protein needed to
perform an in vitro selection by 100 fold — from milligram scale, to 10-100 micrograms. This fact alone
greatly expands the number of proteins that can be targeted by mRNA display, as target production is a
major bottleneck in affinity reagent generation[164].
61
MFEDloadingandwashingwithhighaffinityandmoderateaffinityligands. The flow exper-
iments with the high affinity mRNA-display peptide ligand (E1) established conditions for loading and
washing target-bound beads with the MFED. Using this information, we tested eluting a moderate affinity
ligand (Pep2 K
d
∼ 65 nM) vs. the high affinity E1 peptide, both presented in mRNA display format (Figure
22(A)). The MFED was first used to load and wash radiolabeled [
35
S]-labeled E1 and Pep2 mRNA fusions
separately. Samples were loaded into the MFED at 25 µ L/min and washed at 500 µ L/min. The loading
step represents a compromise—lower flow rates (1 – 10 µ L/min, Figure 21) give more quantitative binding,
but slow the overall function and selection cycle time of the device due to the relatively large volumes in
the loading step (500-1000µ L). Loading at 25µ L/min allows both fairly rapid library/bead binding while
retaining most of the steady-state binding for E1 and Pep2 (52% and 12%, respectively). Also, since the
fraction loaded depends on the formation rate constant (k
on
) these slightly accelerated loading conditions
may provide a small positive bias for complexes with faster on-rates. This approach thus effectively retains
functional library members on the bead bed. Radiolabeled peptide fusion elution and binding were deter-
mined by scintillation counting the flow-through during loading, the buffer used to wash the bead bed,
and, at the end of the experiment, the beads themselves. Figure 22(A) shows the fraction of each peptide
fusion initially introduced that remains bound to the beads as a function of wash time. It also shows the
ratio of E1:Pep2 remaining on the beads.
Washing has a dramatically different effect on peptide ligands with different affinities. For the Pep2
mRNA-peptide fusion, dissociation is well fit by a first-order model (eqn 4.3; Figure 22(A)), with percent
bound (B), wash time (τ w
), dissociation constant (k
off
), minimum percent bound (B
min
), and initial percent
bound (B
0
):
B = (B
0
− B
min
)· e
− k
off
·τ w
+B
min
(4.3)
62
This fit yields a k
off
= 9× 10
− 3
s
− 1
, which is in excellent agreement with results (k
off
= 1.1× 10
− 2
s
− 1
)
obtained through radiolabeled manual competitor-based off-rate measurement (Figure 22(B)). This off rate
is also consistent with there being no rebinding to the beads once dissociation occurs.
(A) (B)
Figure22: (A) MFED loading and washing E1 (red circle) and Pep2 (green circle) mRNA-peptide fusions.
Error bars indicate the standard deviation of three trials. Initial binding for E1 (52%) shows little decay
during washing, whereas Pep2 binding (12 ± 1%) washes out with a first order rate constant k
off
= 9± 2× 10
− 3
s
− 1
. The ratio of E1:Pep2 bound (blue square) plateaus after 1,000 seconds when the Pep2
binding reaches background binding (1.4± 0.3%). (B) Radiolabeled manual competitor-based off-rate
measurement of Pep2 (green triangle). Beads containing Bcl-x
L
were exposed to radiolabeled Pep2, washed,
and resuspended with 100X free Bcl-x
L
. Timepoints were taken after the resuspension in free Bcl-x
L
and the
percent of Pep2 retained on the beads was recorded and normalized such that the initial binding percentage
was 100%. The curve was fit as a single exponential decay using a nonlinear fit in GraphPad Prism. The
k
off
was1.1× 10
− 2
s
− 1
, and a minimum percent bound of 19%.
Peptide fusions with K
d
∼ 100 nM thus fall to background levels after < 15 minutes. On the other hand,
∼ 90% of the E1 peptide fusions are retained during the 30 minute washing step. To achieve an optimal
enrichment of E1 over Pep2, conditions need to be found where the ratio of the stronger and weaker binder
is maximized. During MFED washing, the ratio of E1:Pep2 bound rises rapidly and plateaus after 1,000
seconds.
The individual experiments with E1 and Pep2 peptide fusions demonstrate conditions for efficient
loading and washing for ligands typical in mRNA display selections (K
d
= 100 nM) and those optimized
63
for antibody-like affinity (K
d
≤ 1 nM). More importantly, during washing, the ability to achieve high flow
rates (500 µ L/min) with small bed volumes and short residence times prohibits ligand/target rebinding,
enabling competitor-free off-rate selections.
Testing MFED enrichment of a high affinity ligand (E1) vs. a lower affinity ligand (Pep2).
In a typical manual mRNA display selection, peptide ligands with their attached encoding cDNA bind
immobilized target. Unbound ligands are then washed using a filter, and the remaining ligands along with
their encoding cDNA are collected and amplified by PCR. Our optimization of the washing conditions in the
MFED suggested that the MFED would be more effective than manual selections in enriching high affinity
ligands. The kinetic binding data in Figure 22(A) shows a 4.3-fold enrichment of E1 over Pep2 initially. After
washing, this ratio can be improved to > 40-fold by facilitating an off-rate based enrichment. To quantify
the actual increase of enrichment, we measured the relative abundance of a mixed ligand population before
and after selection using the MFED and manual selection.
Prior to selection, we checked whether PCR would introduce bias during the amplification of E1 and
Pep2 cDNA. PCR bias is present when some templates amplify more efficiently than others, which can lead
to a change in relative abundance that could potentially be misinterpreted as enrichment. Since E1 and
Pep2 are different lengths (126 and 96 bp, respectively), they can be distinguished on an agarose gel. To
ensure that there is no PCR bias between E1 and Pep2, we mixed the two cDNAs such that these species
gave similar band intensities on an agarose gel (Figure 23). This mixture was then serially diluted such that
each sample required a different number of PCR cycles to be amplified to achieve visible band intensities.
The ratio of band intensities remains unchanged with samples PCR amplified for different numbers of
cycles, thereby confirming that there is no PCR bias between E1 and Pep2 (Figure 23).
64
Figure 23: Dilution and PCR of 1:1 mixture of E1 and Pep2 PCR products. The gel image indicates that
after 7 cycles, the band intensities for E1 and Pep2 are comparable, and this continues to be true through
22 cycles. This implies that there is minimal PCR bias.
To demonstrate that selection with the MFED is capable of outperforming a typical manual mRNA
display selection, we used mixtures of E1 and Pep2 as our starting materials. The different lengths of
their coding DNA enabled us to easily quantify enrichment by determining the relative intensities of each
species on an agarose gel.
To test this, we prepared mixtures of E1 and Pep2 at different ratios. E1 and Pep2 were co-translated
at various ratios (E1:Pep2 at 1:2, 1:38, 1:67, 1:135, and 1:292), reverse transcribed, and a sample at each
ratio was then used in either an MFED or traditional manual mRNA display selection. Mixing the two
templates immediately prior to the translation step (rather than an earlier step in the mRNA display cycle,
e.g., PCR, transcription, or ligation) reduces the potential biases for one clone over the other prior to
selection. After reverse transcription, half of each mixture was used for the MFED selection and the other
half was used for manual selection. Splitting each mixture in half directly before selection also reduces the
variation/bias in the samples due to post-translation steps such as dT purification, reverse transcription,
or sample handling.
65
(A) (B)
(C) (D)
Figure 24: Manual versus MFED enrichment of E1 over Pep2. (A) Defined standard mixtures of ligated
E1 and Pep2 were prepared and reverse transcribed. The sample was then amplified by PCR and run on
agarose gels. (B) The band intensity ratios of the PCR amplified standard mixtures were calculated using
Image Studio Lite to build a standard curve. (C) Selection was performed on various starting ratios of
E1:Pep2 (1:2, 1:38, 1:67, 1:135, and 1:292) and the resulting PCR products were run on agarose gels. (D) The
band intensities were measured and converted into molar ratios using the standard curve. MFED selection
consistently outperformed manual selection, averaging 13-fold enrichment compared to 5-fold enrichment
observed with manual selection (* indicates p < 0.05).
In manual selections, each sample was incubated with bead-immobilized Bcl-x
L
in a rotating tube for
an hour and washed over a filter three times, with no off-rate selective pressure added. In MFED selections,
the samples were loaded onto bead-immobilized Bcl-x
L
at a flow rate of 25 µ L/min and then washed at 500
µ L/min for 15 min, which our data above determined to be conditions that would be expected to yield an
off-rate selective pressure. In each case, after the selections were completed, the beads were transferred
into PCR solution and each sample was PCR amplified until the bands were visible on an agarose gel
66
(Figure 24). Image analysis was performed using Image Studio Lite, and molar ratios were extracted from
a standard curve (Figure 24).
In each trial, the MFED outperformed manual selection, averaging a three-fold improvement in en-
richment of E1 over Pep2. This suggests that a selection performed with the MFED and off-rate selective
pressures would likely require fewer cycles of enrichment versus a traditional mRNA display selection,
and therefore could reduce the time it takes to perform a selection from a naïve library. This result shows
that the MFED is superior to traditional mRNA display selection in distinguishing ligands based on their
off-rates and confirms that the device is capable of performing competitor-free, off-rate based selection.
4.5 Conclusions
In our effort to develop automated mRNA display, we designed the MFED as a more economical microflu-
idic alternative to the conventional competitor based off-rate selection technique. The MFED described
here is an innovative device that facilitates a competitor-free, off-rate selection of mRNA display ligands.
We demonstrated that the MFED selection is superior to the manual, non-off-rate-based technique owing
to its continual free-ligand removal mechanism. The device utilizes continuous flow to facilitate ligand
binding and washing with predictable kinetics. With only 100 pmol of target on 5µ L of beads, we demon-
strated that ligands can be efficiently loaded at 25 µ L/min and effectively washed at 500 µ L/min. Both flow
rates are easily attainable with a syringe pump, and the change in flow rate is only 20-fold. Fabrication
of MFED devices is achieved with no specialized lithography equipment and utilizes low cost materials.
The reduction in the amount of protein target that must be expressed and purified to perform off-rate
selections enables access to targets that are difficult or expensive to produce in large quantities. Lastly,
the combination of the lower sample requirements with the higher enrichment achieved with the MFED
device argues that the MFED is an attractive selection improvement to accelerate the discovery of peptide
and protein ligands against novel cancer targets for diagnostic and therapeutic uses.
67
Chapter5
CompatibilityofPopularThree-DimensionalPrintedMicrofluidics
MaterialswithIn-Vitro EnzymaticReactions
Note: This chapter has been published as a journal article. The full reference is: Wan-Zhen Sophie Lin et
al. “Compatibility of Popular Three-Dimensional Printed Microfluidics Materials with In- Vitro Enzymatic
Reactions”. In: ACS Applied Bio Materials 5.2 (Feb. 21, 2022), pp. 818–824. issn: 2576-6422, 2576-6422. doi:
10/gqjqcw. url: https://pubs.acs.org/doi/10.1021/acsabm.1c01180 (visited on 07/23/2022)
5.1 Abstract
3D printed microfluidics offer several advantages over conventional planar microfabrication techniques
including fabrication of 3D microstructures, rapid prototyping, and inertness. While 3D printed materials
have been studied for their biocompatibility in cell and tissue culture applications, their compatibility for
in vitro biochemistry and molecular biology has not been systematically investigated. Here, we evaluate
the compatibility of several common enzymatic reactions in the context of 3D-printed microfluidics: 1)
polymerase chain reaction (PCR), 2) T7 in vitro transcription, 3) mammalian in vitro translation, and 4)
reverse transcription. Surprisingly, all the materials tested significantly inhibit one or more of these in- vitro
enzymatic reactions. Inclusion of BSA mitigates only some of these inhibitory effects. Overall, inhibition
68
appears to be due to a combination of the surface properties of the resins as well as soluble components
(leachate) originating in the matrix.
5.2 Motivation
Microfluidics is an important tool in areas of biochemistry to miniaturize bulky and costly laboratory pro-
cesses. By confining reactions and processes within micron-sized channels, microfluidic systems offer high
portability, minimal reagent consumption, and sophisticated process integration at low costs[148, 101, 70,
84, 138, 39, 96]. For mass production, microfluidic devices can be manufactured by high-throughput tech-
niques such as injection molding[107, 50, 178] and hot embossing[11, 94, 78, 89, 29, 144]. For prototyping
in research settings, however, microfluidic devices are typically fabricated in small batches through mold-
ing elastomers or thermoplastics[163, 143, 137, 48, 26]. The fabrication process of microfluidic molds are
tedious and costly, and the resulting devices are limited to two-dimensional (2D) structures with cumber-
some user interfaces[3]. Because of these limitations, more and more microfluidic researchers have turned
to three-dimensional (3D) printing for microfabrication[39, 163, 3, 151, 166, 63, 2, 60, 16].
In 3D-printed microfluidics, the devices are modeled using computer aided design software and then
printed additively layer-by-layer. Among all 3D printing techniques, stereolithography (SLA) is the most
widely used for microfluidics because of its high printing resolution and capability of fabricating closed
microfluidic channels without additional assembly steps[166, 52, 124]. SLA printers selectively polymerize
liquid photopolymer resins using a focused laser light source[69]. The additive layers of polymerized resin
form microchannel walls that are filled with liquid resin, which is flushed out of the microfluidic channels
afterwards[166, 63]. With previously unattainable simplicity, the technique opens the possibility to single-
step microfabrication, overcomes the 2D limitations of conventional fabrication methods, and incorporates
industry-standard user interfaces that are commercialization-ready[3, 166, 60, 14].
69
Despite its unique advantages, 3D-printed microfluidics have not been readily adopted by the biochem-
istry community and recent work has been limited to proof-of-concept studies[166]. This is in part due to
the stringent material compatibility requirements for biochemical reactions. In microfluidic channels, the
interaction between biological molecules and the material of the channel wall is especially pronounced be-
cause of the high surface-area-to-volume ratio[163, 54]. Today, most microfluidic devices for biochemical
reactions are fabricated by traditional polydimethylsiloxane (PDMS) molding despite its tedious fabrica-
tion process and poor user interfaces[96, 109, 43, 134]. Many microfluidic researchers are reluctant to part
with PDMS molding because of the optimal material properties: PDMS is optically clear, chemically inert,
biocompatible, and easy to mold[133, 45]. The dissemination of 3D printed microfluidics is therefore chal-
lenged by the compatibility of 3D-printable materials with biochemical reactions[163, 16, 52, 124, 54, 135,
51, 23].
Previous investigations on 3D printable photopolymers for biological applications focused on cell cul-
ture compatibility[23, 183, 117, 154], but little has been done to evaluate their inhibitory effects on in- vitro
enzymatic reactions. In-vitro enzymatic reactions, such as PCR, are essential for lab-on-a-chip applications,
where the transition from traditional microfabrication techniques to 3D printing has been longed for. To
help break the barriers of microfluidics application in the biochemistry community, we evaluated some
of the most popular SLA-printed materials for their compatibility with polymerase chain reaction (PCR),
transcription, translation, and reverse transcription. We designed a 3D printed cone that sits inside con-
ventional conical microcentrifuge tubes to mimic the surface contact of the reactions within microfluidic
channels.
Including BSA in PCR solutions as a blocking agent for materials that adsorbs the polymerase is well
documented[131, 156, 85, 169]. For SLA-printed materials, it was speculated that in-vitro enzymatic reac-
tions may be inhibited by the substances, such as uncured resins, leaching out from the printed parts in
70
addition to the effects of polymerase adsorption[87]. By tuning the concentration of BSA for each reaction,
we evaluated current limitations of SLA-printed materials for microfluidics in- vitro enzymatic reactions.
5.3 MaterialsandMethods
5.3.1 3DPrinting
To evaluate the compatibility of SLA printed parts with in-vitro biological reactions, we prepared 3D
printed parts of three common SLA materials for microfluidics: WaterShed (Somos WaterShed XC 11122,
DSM), Pro3dure GR-1 (Pro3dure Medical GmbH, Dortmund, Germany), and a PEGDA-based resin prepared
in-house[51]. For surface treatment, a commercial Teflon coating (Teflon Non-Stick Dry-Film Lubricant,
Dupont, USA) was applied onto parts printed with Pro3dure GR-1. The 3D printed parts were designed
with a computer-aided design software (Autodesk Fusion 360). Each part was a designed in cone such
that, when placed with 50µ L liquid sample in a 500µ L polypropylene microcentrifuge tube (Eppendorf,
Germany), the liquid is in contact with the 3D printed parts at an equivalent surface-area-to-volume ratio
to a 1.5 mm-diameter channel (Figure 25).
Figure25: Illustration of a 3D-printed cone placed in a microcentrifuge tube.
71
5.3.2 PCR
PCR was performed using DreamTaq™ Hot Start DNA Polymerase from Thermo Scientific ™. We used the
template of a previously identified Bcl-x
L
ligand, labeled E1 (amino acid sequence MIETITIYNYKKAAD-
HFSMSM), to test PCR. The resulting product was used to test transcription, translation, and reverse tran-
scription as described in our previous work[39]. The PCR reactions were performed in the supplied buffer
with 2 mM dNTPs, 100 pM of template DNA, 1 µ M primers, and various concentrations of BSA. Stan-
dard thermocycling (94°C - 53°C - 72°C cycling at 30 seconds per step) was performed for 10 cycles using
polypropylene tubes with or without additional 3D-printed parts in the tube. The samples were run on an
agarose gel.
5.3.3 Transcription
Transcription was performed using T7 RNA polymerase that was expressed and purified in house using
a his tag. Reactions were performed in 80 mM HEPES pH = 7.4, 2 mM Spermidine, 40 mM DTT, 25 mM
MgCl2, 5 mM each NTP, 50 nM of template DNA, and various concentrations of BSA. Incubation was
performed on a heating block set to 37°C for 1 hour in polypropylene tubes with or without additional
3D-printed parts in the tube. The samples were run on a acrylamide urea gel.
5.3.4 Translation
Translation was performed as previously described by Jackson and Hunt using rabbit reticulocyte lysate
(RRL)[72]. The unlabeled methionine was replaced with
35
S methionine (Perkin Elmer) for radiolabel
the resulting peptides. Reactions were performed in 40% treated RRL, 20 mM HEPES pH = 7.6, 0.5 mM
Mg(OAc)
2
, 8 mM creatine phosphate, 2 mM DTT, 25 µ M each amino acid, 400 nM of template ligated
mRNA, and various concentrations of BSA. Incubation was performed on a heating block set to 30°C for 1
hour in polypropylene tubes with or without additional 3D-printed parts in the tube and KCl and MgCl
2
72
salts were added to a final concentration of 470 mM and 60 mM respectively to facilitate fusion formation.
Fusions were purified from the translation mixture using dT beads (streptavidin agarose beads, PierceTM,
and biotinylated dT25 oligo, IDT), and the beads were counted on a scintillation counter (Beckman LS
6500).
5.3.5 ReverseTranscription
Reverse transcription was performed with Superscript IV in 50 mM Tris-HCl (pH=8.3), 75 mM KCl, 3 mM
MgCl
2
, 10 mM DTT, 500µ M each dNTP, and 1 mM 3’ primer on 200 nM of template mRNA. The reaction
was incubated at 42°C for 1 hour to facilitate reverse transcription, then further incubated at 65°C for 15
min to inactivate Superscript IV. The sample was PCR amplified and then ran on an agarose gel to evaluate
the yield from reverse transcription.
5.4 ResultsandDiscussion
PCR. In an attempt to optimize PCR yield with BSA in the presence of 3D printed parts, we first evaluated
how PCR reaction yield varies with BSA concentration (Figure 26(A)). As shown in the figure, PCR yield
was consistent among 0 3 wt% BSA samples. At concentrations above 0.6 wt% BSA, the samples become
gel-like after the reaction due to BSA aggregation.51 For microfluidics applications, the increased viscos-
ity can cause the microchannels to clog. Therefore, we proceeded to evaluate the effects of 3D printed
parts on PCR with 0.4 wt% BSA, which is the highest BSA concentration deemed suitable for microfluidic
applications. We performed two sets of experiments: with and without BSA (0.4 wt%). In each set, PCR
reactions were performed in tubes each containing 3D printed cones of different materials. As shown in
Figure 26(B), samples with 0.4 wt% BSA resulted in significantly higher PCR yields with an average of 20%
increase from the sample without BSA or parts.
73
(A) (B)
Figure 26: (A) PCR yield (normalized to no BSA, no parts sample) at various BSA wt%. 0.4 wt% BSA
(highlighted in grey) was chosen for the following experiments. (B) PCR yield (normalized to no BSA, no
parts sample) when incubated with different 3D printed materials, performed with and without BSA (0.4
wt%). BSA significantly improved PCR yields when 3D printed parts were present (p < 0.05). Gel images
shown in Figure 27.
(A)
(B)
Figure27: PCR Results. (A) A gel image showing PCR outcomes at different wt% BSA concentrations. (B)
A gel image showing PCR outcomes when incubated with 3D-printed parts with or without BSA. Ladder:
100bp DNA Ladder, New England BioLabs. Electrophoresis Buffer: 1X SB.
74
Transcription. Our early attempts to perform T7 transcription in 3D printed microfluidic channels
failed to produce product. While PEG is a more common blocking agent for T7 transcription than BSA,
the high viscosity of PEG is unfavorable for microfluidics. To address how BSA might recover activity, we
first evaluated how transcription yield varies at different BSA concentrations. As shown in Figure 28(B),
transcription yield decreased drastically with increased BSA concentration above 0.9 wt%. We suspect
that this may be due to BSA competing with the polymerase. Additionally, since we observed sample
gelling above 0.6 wt%, we decided to proceed with 0.4 wt% BSA. As shown in Figure 28(B), transcription
samples that were incubated with 3D printed materials did not result in observable yield. With 0.4 wt%
BSA, samples that were incubated with WaterShed XC, Pro3dure GR-1, and Teflon-coated Pro3dure GR-1
showed significant improvements in yield. The samples with parts printed with the PEGDA-based SLA
resin failed both with and without BSA. The results suggest that experiments involving T7 polymerase
production of RNA may not work well in any of the tested microfluidic materials without taking additional
steps.
(A) (B)
Figure28: (A) Transcription yield (normalized to no BSA, no parts sample) at various BSA wt%. 0.4 wt%
BSA (highlighted in grey) was chosen for the following experiments. (B) Transcription yield (normalized to
no BSA, no parts sample) when incubated with different 3D printed materials, performed with and without
BSA (0.4 wt%). BSA significantly improved the transcription yield when 3D printed parts were present (p
< 0.05). The yields of printed parts without BSA and PEGDA-based resin with 0.4 wt% BSA were below
the detection limit (Figure 29).
75
(A)
(B)
(C)
Figure 29: Transcription Results. (A) A gel image showing transcription outcomes at different wt% BSA
concentrations. (B) A gel image showing transcription outcomes when incubated with 3D-printed parts
with or without BSA. (C) A photo of the samples collected in Eppendorf tubes showing gelation at high
BSA concentrations. Electrophoresis Buffer: 1X TBE.
Translation. As shown in Figure 30(A), translation yield decreases slightly as BSA concentration in-
creased. While the samples remained liquid at all BSA concentrations tested, the viscosity of the sample
increased with the BSA content, which would increase the risk of channel fouling and clogging in mi-
crofluidics settings. Therefore, to mitigate the inhibiting effects of 3D printed parts without compromising
the microfluidic flow performance, we decided to proceed with testing the compatibility of 3D printed
materials with 0.6 wt% BSA, a slightly higher concentration than what was used for PCR and transcription
(0.4 wt%). The results showed no significant improvement (based on Student’s t-test) of BSA on translation
yields across different 3D printed materials (Figure 30(B)).
76
(A) (B)
Figure30: (A) Translation yield (normalized to no BSA, no parts sample) at various BSA wt%. 0.6 wt% BSA
(highlighted in grey) was chosen for the subsequent experiments. (B) Translation yield (normalized to no
BSA, no parts sample) when incubated with different 3D printed materials, performed with and without
BSA (0.6 wt%). BSA did not significantly improve translation yield when 3D printed parts were present.
Reverse Transcription. As shown in Figure 31(A), reverse transcription yield lowered as BSA con-
centration increased, and all samples remained liquid after the reaction. Based on the results, we decided
to proceed with 0.6 wt% BSA as for the translation experiments so as to not compromise the reaction yield
too much. As shown in Figure 31(B), 0.6 wt% BSA significantly improved the reverse transcription yield
when incubated with all 3D printed materials.
77
(A) (B)
Figure 31: (A) Reverse transcription yield (normalized to no BSA, no parts sample) at various BSA wt%.
0.4 wt% BSA (highlighted in grey) was chosen for the subsequent experiments. (B) Transcription yield
(normalized to no BSA, no parts sample) when incubated with different 3D printed materials, performed
with and without BSA (0.6 wt%). BSA significantly improved the reverse transcription yield when 3D
printed parts were present (p< 0.05). The yields of PEGDA-based resin without BSA and Pro3dure GR-1
without BSA were below the detection limit (Figure 32).
78
(A)
(B)
Figure32: Reverse Transcription Results. (A) A gel image showing reverse transcription outcomes at dif-
ferent wt% BSA concentrations. (B) A gel image showing reverse transcription outcomes when incubated
with 3D-printed parts with or without BSA. Ladder: 100bp DNA Ladder, New England BioLabs. Elec-
trophoresis Buffer: 1X SB.
TranscriptionusingWaterwithLeachate. To better understand the proportional impact of adsorp-
tion versus reaction-inhibiting substances being leached out from the parts, we performed transcription
experiments with water that was previously incubated for 12 hours at 37°C with each 3D printed part.
Note that all parts were treated in a post-print UV curing chamber, minimizing the amount of unreacted
monomer present. During the extended soaking, water-soluble substances leached out. The leachate water
collected after extended soaking was diluted 5x in the transcription sample to evaluate its impact on tran-
scription. We chose to test with transcription because it is the easiest to perform among all reactions and
was very sensitive to both BSA and the type of substrate material as shown in Figure 28. The amount of
water added to each reaction resulted in a 50% dilution of the leachate. In comparison to Figure 28, Figure
33 shows that the reaction yield was slightly improved in the WaterShed XC and Pro3dure GR-1 trials,
79
although still lower than the sample without BSA or parts. This suggests that the inhibiting substances
being leached out from the 3D printed parts significantly hindered the reaction.
Figure 33: Transcription yield (normalized to no BSA, no parts sample) when incubated with parts vs
with leachate isolated from parts. The yields from PEGDA-based resin and Teflon-coated Pro3dure GR-1
were below the detection limit (Figure 34).
Figure34: Transcription using Water with leachate. The gel image shows the transcription outcome with
leachates collected from different materials. Electrophoresis Buffer: 1X TBE.
The results shown above indicate that PCR, transcription, and reverse transcription can be improved
with the addition of BSA. The addition of BSA did not impact translation yields significantly, perhaps
because the reaction uses lysate, which already includes additional proteins that alleviate the inhibiting
effects from the leachates. Among all materials tested, we found that WaterShed XC performed the best
overall. The PEGDA-based material performed the worst, with non-observable transcription yield even
with the addition of BSA. All 3D printed materials assessed in this study significantly inhibits in- vitro
80
enzymatic reactions. This may be due to enzymes being adsorbed onto the 3D printed surfaces and/or
inhibiting substances being leached out of the 3D printed materials. When comparing the Pro3dure GR-1
with and without Teflon coating, no significant difference in reaction yields were observed, which suggests
that inhibiting substances leaching out of the material during the reaction plays an important role in the
low reaction yields.
The transcription experiments performed with only the leachate but not the parts shows low reaction
yields relative to the no BSA, no parts samples (Figure 33). This results further confirmed that the leachate
from materials have significant inhibiting effects on the reactions, possibly due to their interaction with the
polymerase. By comparing the transcription with leachate results (Figure 33) with the transcription with
parts results (Figure 28(B)), we can see that the reaction yield was even lower when incubated with the
parts. This indicates that aside from the leachate, transcription was also inhibited by interactions between
the polymerase and the parts. Furthermore, the transcription yields with parts and with BSA (Figure 28)
were higher than the yields obtained from transcription with leachate (Figure 33), indicating that BSA
mitigates the interaction of the polymerase and both the leachate and the materials. We speculate that
the mitigating effect of BSA is caused by BSA competing with the polymerase for interactions with the
materials.
In order for 3D-printed microfluidic parts to be usable for in- vitro enzymatic reacssoawttions, fur-
ther materials development is needed. Post-print channel modifications with covalent fluorosilane and
dimethylsiloxane coatings have been shown to increase PCR yields by minimizing the interactions be-
tween the enzymes and 3D printed materials and inhibiting leaching[161]. New resin formulations for
3D-printed parts that are compatible with in-vitro enzymatic reactions would require elaborate screening
of each ingredient. We speculate that the limited photoinitiator selection currently available for transpar-
ent prints is incompatible for in-vitro enzymatic reactions and that new photoinitiators should be explored
to overcome the challenge. Although rapid prototyping techniques that do not involve photocurable resins
81
(e.g. fused deposition modeling[160, 149, 123, 28] and 3D CNC milling[115]) may provide a wider range
of materials selection, studies of these materials have also shown decreased enzymatic activities[123, 115].
This suggests the need for the development of dedicated materials specifically designed for biochemical
applications.
5.5 Conclusion
In this study, we evaluated 3D-printed materials that are popular in microfluidics for their compatibility
with in-vitro enzymatic reactions. In the attempt to mitigate the inhibiting effects of the materials to
the reactions, we added BSA to the reactions and saw improvements in PCR, transcription, and reverse
transcription, but not in translation. We suspect the lysate used in translation already contains proteins
that can compete with the polymerase for interactions with the printed materials, and therefore adding
more BSA did not further improve the reaction yield.
Based on our findings, WaterShed XC is the most compatible with in- vitro enzymatic reactions, al-
though the reactions were nonetheless significantly inhibited. Since Teflon-coated Pro3dure GR-1 did not
show improvement over the non-coated material, we suspect that the inhibiting effects are most likely due
to leachate from the materials in addition to adsorption of the enzymes to material surfaces. This spec-
ulation was later confirmed in our transcription experiments incubated with only the leachate, without
the 3D-printed parts. We conclude that both the leachate and the materials caused significant inhibition
to in-vitro enzymatic reactions, and that BSA appears essential to mitigate the inhibiting effects, although
the improvements were marginal. The finding is especially concerning as the incompatibly of 3D printed
materials with in-vitro enzymatic reaction could be a major barrier for applying 3d printing technology
to biomicrofluidics. The development of new SLA materials for biomicrofluidics should focus on reducing
leachate and minimizing polymerase interaction.
82
Chapter6
ConclusionsandOutlook
The studies presented in this dissertation emphasize the utility of microfluidics for biopharmaceutical ap-
plications, including two devices for nanomedicine fabrication and a continuous flow device that facilitates
affinity ligand screening. Microfluidic technology provides unprecedented advantages over conventional
processing techniques because of its intricate nature, which enables extraordinary control over mass, en-
ergy, and momentum transfer within localized micro-environment. Our initial work on liposome produc-
tion employed a planar microfluidic hydrodynamic focusing device, which generated drug-loaded lipo-
somes with tunable particle characteristics. Despite current efforts in developing microfluidic platforms,
very few could practically replace their conventional counterparts. This is mainly because of 1) the cum-
bersome microfluidic device fabrication process, 2) the poor fluidic interfaces in traditional PDMS-based
microfluidic devices, and 3) difficult fluid operations with unreliable microfluidic valves and pumps. Our
approach to address these issues is by transforming microfluidics into 3D printable formats. Compared to
elastomer micromolding, microfluidic device fabrication via 3D printing is much easier and faster, and the
devices can be built with user-friendly interfaces that are compatible with commercially available fluidic
connectors. This enables easy integration of microfluidic devices with fluid automation equipment (e.g.
solenoid valves and syringe pumps) that are cost-effective, reliable, robust, and often readily available in
scientific laboratories. We leveraged the ability to create 3D microfluidic channels and designed a powerful
83
4-way sheath flow device that facilitates hydrodynamic focusing at high flow rates. The 4-way sheath flow
design has the potential to revolutionize nanoparticle fabrication. We anticipate its application to extend
beyond lipid nanoparticle synthesis; to find applications in various nanoparticle synthesis that benefit from
the precise control over multi-phase fluid mixing, such as complex emulsions. For target-directed ligand
screening, we built a microfluidic enrichment device that utilizes a continuous-flow to facilitate off-rate
based selection. The study suggests the potential of automating sophisticated, labor-intensive biochemical
processes using microfluidics. However, our evaluation on popular, commercially-available photo-curable
resins showed that most of the resins currently used in stereolithography printing have low compatibility
with in vitro enzymatic reactions due to toxic leachates and surface adsorption, impeding their applica-
tions for biochemical processes. This field is still largely unexplored, and should be further investigated to
answer the emerging questions about the utility of 3D printed microfluidics for biopharmaceutical devel-
opment processes that involve bio-enzymatic reactions.
84
Bibliography
[1] Aaron C Anselmo and Samir Mitragotri. “Nanoparticles in the clinic: An update”. In:
Bioengineering & Translational Medicine 4.3 (2019). Publisher: Wiley Online Library, e10143.
[2] Anthony K. Au, Nirveek Bhattacharjee, Lisa F. Horowitz, Tim C. Chang, and Albert Folch.
“3D-printed microfluidic automation”. In: Lab on a Chip 15.8 (2015). Publisher: Royal Society of
Chemistry, pp. 1934–1941.issn: 14730189.doi: 10/f7d3ck.
[3] Anthony K. Au, Wilson Huynh, Lisa F. Horowitz, and Albert Folch. “3D-Printed Microfluidics”.
In: Angewandte Chemie - International Edition 55.12 (2016), pp. 3862–3881.issn: 15213773.doi:
10.1002/anie.201504382.
[4] Anthony K. Au, Wonjae Lee, and Albert Folch. “Mail-order microfluidics: evaluation of
stereolithography for the production of microfluidic devices”. In: Lab Chip 14.7 (2014),
pp. 1294–1301.issn: 1473-0197, 1473-0189.doi: 10/f5zv6z. (Visited on 07/13/2022).
[5] Lindsey R. Baden, Hana M. El Sahly, Brandon Essink, Karen Kotloff, Sharon Frey, Rick Novak,
David Diemert, Stephen A. Spector, Nadine Rouphael, C. Buddy Creech, John McGettigan,
Shishir Khetan, Nathan Segall, Joel Solis, Adam Brosz, Carlos Fierro, Howard Schwartz,
Kathleen Neuzil, Lawrence Corey, Peter Gilbert, Holly Janes, Dean Follmann, Mary Marovich,
John Mascola, Laura Polakowski, Julie Ledgerwood, Barney S. Graham, Hamilton Bennett,
Rolando Pajon, Conor Knightly, Brett Leav, Weiping Deng, Honghong Zhou, Shu Han,
Melanie Ivarsson, Jacqueline Miller, and Tal Zaks. “Efficacy and Safety of the mRNA-1273
SARS-CoV-2 Vaccine”. In: New England Journal of Medicine 384.5 (Feb. 4, 2021), pp. 403–416.issn:
0028-4793, 1533-4406.doi: 10/ghrg8m. (Visited on 07/12/2022).
[6] Kapil Bahl, Joe J. Senn, Olga Yuzhakov, Alex Bulychev, Luis A. Brito, Kimberly J. Hassett,
Michael E. Laska, Mike Smith, Örn Almarsson, James Thompson, Amilcar (Mick) Ribeiro,
Mike Watson, Tal Zaks, and Giuseppe Ciaramella. “Preclinical and Clinical Demonstration of
Immunogenicity by mRNA Vaccines against H10N8 and H7N9 Influenza Viruses”. In: Molecular
Therapy 25.6 (June 7, 2017), pp. 1316–1327.issn: 1525-0016.doi: 10/gk5nhq. (Visited on
07/11/2022).
[7] Amrita Banerjee, Jianping Qi, Rohan Gogoi, Jessica Wong, and Samir Mitragotri. “Role of
nanoparticle size, shape and surface chemistry in oral drug delivery”. In: Journal of Controlled
Release 238 (2016). Publisher: Elsevier, pp. 176–185.doi: 10.1016/j.jconrel.2016.07.051.
85
[8] AD Bangham, Malcolm M Standish, and Jeff C Watkins. “Diffusion of univalent ions across the
lamellae of swollen phospholipids”. In: Journal of molecular biology 13.1 (1965). Publisher:
Elsevier, 238–IN27.doi: 10/ch349t.
[9] Yechezkel Barenholz. “Relevancy of drug loading to liposomal formulation therapeutic efficacy”.
In: Journal of liposome research 13.1 (2003). Publisher: Taylor & Francis, pp. 1–8.doi: 10/c2drwz.
[10] Shmuel Batzri and Edward D Korn. “Single bilayer liposomes prepared without sonication”. In:
Biochimica et Biophysica Acta (BBA)-Biomembranes 298.4 (1973). Publisher: Elsevier,
pp. 1015–1019.doi: 10/bmk88r.
[11] Holger Becker and Ulf Heim. “Hot embossing as a method for the fabrication of polymer high
aspect ratio structures”. In: Sensors and Actuators, A: Physical 83.1 (May 22, 2000). Publisher:
Elsevier Sequoia SA, pp. 130–135.issn: 09244247.doi: 10/c669gw.
[12] David J Beebe, Glennys A Mensing, and Glenn M Walker. “Physics and applications of
microfluidics in biology”. In: Annual review of biomedical engineering 4.1 (2002). Publisher:
Annual Reviews 4139 El Camino Way, PO Box 10139, Palo Alto, CA 94303-0139, USA,
pp. 261–286.doi: 10/ckwstv.
[13] Nathan M Belliveau, Jens Huft, Paulo JC Lin, Sam Chen, Alex KK Leung, Timothy J Leaver,
Andre W Wild, Justin B Lee, Robert J Taylor, Ying K Tam, Carl L Hansen, and Pieter R Cullis.
“Microfluidic Synthesis of Highly Potent Limit-size Lipid Nanoparticles for In Vivo Delivery of
siRNA”. In: Molecular Therapy - Nucleic Acids 1 (2012), e37.issn: 21622531.doi:
10.1038/mtna.2012.28. (Visited on 07/10/2022).
[14] Krisna C Bhargava, Bryant Thompson, and Noah Malmstadt. “Discrete elements for 3D
microfluidics”. In: Proceedings of the National Academy of Sciences 111.42 (2014). Publisher:
National Acad Sciences, pp. 15013–15018.doi: 10/f6kpp4.
[15] Krisna C. Bhargava, Roya Ermagan, Bryant Thompson, Andrew Friedman, and Noah Malmstadt.
“Modular, discrete micromixer elements fabricated by 3D printing”. In: Micromachines 8.5 (2017).
Publisher: MDPI AG, p. 137.issn: 2072666X.doi: 10/gqjqcp.
[16] Nirveek Bhattacharjee, Arturo Urrios, Shawn Kang, and Albert Folch. “The upcoming
3D-printing revolution in microfluidics”. In: Lab on a Chip 16.10 (2016). Publisher: Royal Society
of Chemistry, pp. 1720–1742.issn: 14730189.doi: 10/gf4dj6.
[17] Eric T Boder, Katarina S Midelfort, and K Dane Wittrup. “Directed evolution of antibody
fragments with monovalent femtomolar antigen-binding affinity”. In: Proceedings of the National
Academy of Sciences 97.20 (2000). Publisher: National Acad Sciences, pp. 10701–10705.doi:
10/b5n7t4.
[18] Eric T Boder and K Dane Wittrup. “Optimal screening of surface-displayed polypeptide libraries”.
In: Biotechnology progress 14.1 (1998). Publisher: Wiley Online Library, pp. 55–62.doi:
10.1021/bp970144q.
86
[19] Eric T Boder and K Dane Wittrup. “Yeast surface display for screening combinatorial polypeptide
libraries”. In: Nature biotechnology 15.6 (1997). Publisher: Nature Publishing Group, pp. 553–557.
doi: 10/b8fjrq.
[20] Andrew Bradbury and Andreas Plückthun. “Reproducibility: Standardize antibodies used in
research”. In: Nature 518.7537 (2015). Publisher: Nature Publishing Group, pp. 27–29.doi:
10/gkfkdf.
[21] M. Budai and M. Szógyi. “Liposomes as drug carrier systems. Preparation, classification and
therapeutic advantages of liposomes”. In: Acta Pharmaceutica Hungarica 71.1 (2001), pp. 114–118.
issn: 0001-6659.
[22] Dario Carugo, Elisabetta Bottaro, Joshua Owen, Eleanor Stride, and Claudio Nastruzzi. “Liposome
production by microfluidics: potential and limiting factors”. In: Scientific reports 6 (2016).
Publisher: Nature Publishing Group, p. 25876.doi: 10/f8m548.
[23] Megan Carve and Donald Wlodkowic. “3D-printed chips: Compatibility of additive
manufacturing photopolymeric substrata with biological applications”. In: Micromachines 9.2
(2018).issn: 2072666X.doi: 10/gg8h44.
[24] Mehmet Cetin, William E Evenson, Garrett G Gross, Farzad Jalali-Yazdi, Daniel Krieger,
Don Arnold, Terry T Takahashi, and Richard W Roberts. “RasIns: genetically encoded intrabodies
of activated Ras proteins”. In: Journal of molecular biology 429.4 (2017). Publisher: Elsevier,
pp. 562–573.doi: 10/f9stsv.
[25] Tim C. Chang, Andrei M. Mikheev, Wilson Huynh, Raymond J. Monnat, Robert C. Rostomily, and
Albert Folch. “Parallel microfluidic chemosensitivity testing on individual slice cultures”. In: Lab
Chip 14.23 (2014), pp. 4540–4551.issn: 1473-0197, 1473-0189.doi: 10/f66wzw. (Visited on
07/11/2022).
[26] C. F. Chen, J. Liu, L. P. Hromada, C. W. Tsao, C. C. Chang, and D. L. DeVoe. “High-pressure needle
interface for thermoplastic microfluidics”. In: Lab on a Chip 9.1 (Jan. 7, 2009). Publisher: Royal
Society of Chemistry, pp. 50–55.issn: 14730189.doi: 10.1039/b812812j.
[27] Helena N Chia and Benjamin M Wu. “Recent advances in 3D printing of biomaterials”. In: Journal
of biological engineering 9.1 (2015). Publisher: BioMed Central, pp. 1–14.doi: 10/gctr3q.
[28] Dagmar Chudobova, Kristyna Cihalova, Sylvie Skalickova, Jan Zitka,
Miguel Angel Merlos Rodrigo, Vedran Milosavljevic, David Hynek, Pavel Kopel, Radek Vesely,
Vojtech Adam, and Rene Kizek. “3D-printed chip for detection of methicillin-resistant
Staphylococcus aureus labeled with gold nanoparticles”. In: Electrophoresis 36.3 (2015),
pp. 457–466.issn: 15222683.doi: 10/f22x96.
[29] Ferah Çoğun, Ender Yıldırım, and M. A. Sahir Arikan. “Investigation on replication of microfluidic
channels by hot embossing”. In: Materials and Manufacturing Processes 32.16 (Dec. 10, 2017).
Publisher: Taylor and Francis Inc., pp. 1838–1844.issn: 15322475.doi: 10/gp6hf9.
87
[30] Kizzmekia S. Corbett, Darin K. Edwards, Sarah R. Leist, Olubukola M. Abiona,
Seyhan Boyoglu-Barnum, Rebecca A. Gillespie, Sunny Himansu, Alexandra Schäfer,
Cynthia T. Ziwawo, Anthony T. DiPiazza, Kenneth H. Dinnon, Sayda M. Elbashir,
Christine A. Shaw, Angela Woods, Ethan J. Fritch, David R. Martinez, Kevin W. Bock,
Mahnaz Minai, Bianca M. Nagata, Geoffrey B. Hutchinson, Kai Wu, Carole Henry, Kapil Bahl,
Dario Garcia-Dominguez, LingZhi Ma, Isabella Renzi, Wing-Pui Kong, Stephen D. Schmidt,
Lingshu Wang, Yi Zhang, Emily Phung, Lauren A. Chang, Rebecca J. Loomis, Nedim Emil Altaras,
Elisabeth Narayanan, Mihir Metkar, Vlad Presnyak, Cuiping Liu, Mark K. Louder, Wei Shi,
Kwanyee Leung, Eun Sung Yang, Ande West, Kendra L. Gully, Laura J. Stevens,
Nianshuang Wang, Daniel Wrapp, Nicole A. Doria-Rose, Guillaume Stewart-Jones,
Hamilton Bennett, Gabriela S. Alvarado, Martha C. Nason, Tracy J. Ruckwardt, Jason S. McLellan,
Mark R. Denison, James D. Chappell, Ian N. Moore, Kaitlyn M. Morabito, John R. Mascola,
Ralph S. Baric, Andrea Carfi, and Barney S. Graham. “SARS-CoV-2 mRNA vaccine design enabled
by prototype pathogen preparedness”. In: Nature 586.7830 (Oct. 22, 2020), pp. 567–571.issn:
0028-0836, 1476-4687.doi: 10.1038/s41586-020-2622-0. (Visited on 07/11/2022).
[31] Randy Crawford, Belma Dogdas, Edward Keough, R. Matthew Haas, Wickliffe Wepukhulu,
Steven Krotzer, Paul A. Burke, Laura Sepp-Lorenzino, Ansuman Bagchi, and Bonnie J. Howell.
“Analysis of lipid nanoparticles by Cryo-EM for characterizing siRNA delivery vehicles”. In:
International Journal of Pharmaceutics 403.1 (Jan. 2011), pp. 237–244.issn: 03785173.doi:
10/cc2sc7. (Visited on 07/09/2022).
[32] S Scott Crump. Apparatus and method for creating three-dimensional objects. 1992.
[33] Ping Cui and Sicen Wang. “Application of microfluidic chip technology in pharmaceutical
analysis: A review”. In: Journal of Pharmaceutical Analysis 9.4 (2019). Publisher: Elsevier,
pp. 238–247.doi: 10/gh7wxn.
[34] Kellye Cung, Russell L Slater, Yue Cui, Sharon E Jones, Habib Ahmad, Rajesh R Naik, and
Michael C McAlpine. “Rapid, multiplexed microfluidic phage display”. In: Lab on a Chip 12.3
(2012). Publisher: The Royal Society of Chemistry, pp. 562–565.doi: 10/d45mg8.
[35] B-J De Gans, Paul C Duineveld, and Ulrich S Schubert. “Inkjet printing of polymers: state of the
art and future developments”. In:Advancedmaterials 16.3 (2004). Publisher: Wiley Online Library,
pp. 203–213.doi: 10.1002/adma.200300385.
[36] Rupak Doshi, Beverly R. Chen, Cecile Rose T. Vibat, Norman Huang, Chang-Wook Lee, and
Geoffrey Chang. “In vitro nanobody discovery for integral membrane protein targets”. In:
Scientific Reports 4.1 (May 2015), p. 6760.issn: 2045-2322.doi: 10.1038/srep06760. (Visited on
07/24/2022).
[37] Erik C Dreaden, Lauren A Austin, Megan A Mackey, and Mostafa A El-Sayed. “Size matters: gold
nanoparticles in targeted cancer drug delivery”. In: Therapeutic delivery 3.4 (2012). Publisher:
Future Science, pp. 457–478.doi: 10/gqjqdd.
[38] Dwaine F. Emerich and Christopher G. Thanos. “The pinpoint promise of nanoparticle-based
drug delivery and molecular diagnosis”. In: Biomolecular Engineering 23.4 (Sept. 2006),
pp. 171–184.issn: 1389-0344.doi: 10/b3njf6.
88
[39] William E. Evenson, Wan Zhen Sophie Lin, Kenmond Pang, Alexander T. Czaja,
Farzad Jalali-Yazdi, Terry T. Takahashi, Noah Malmstadt, and Richard W. Roberts. “Enabling
Flow-Based Kinetic Off-Rate Selections Using a Microfluidic Enrichment Device”. In: Analytical
Chemistry 92.15 (Aug. 4, 2020). Publisher: American Chemical Society, pp. 10218–10222.issn:
15206882.doi: 10/gqjqcq.
[40] Yulia Eygeris, Siddharth Patel, Antony Jozic, and Gaurav Sahay. “Deconvoluting Lipid
Nanoparticle Structure for Messenger RNA Delivery”. en. In: Nano Letters 20.6 (June 2020),
pp. 4543–4549.issn: 1530-6984, 1530-6992.doi: 10.1021/acs.nanolett.0c01386. (Visited on
08/26/2022).
[41] Susanna Fafenrot, Nils Grimmelsmann, Martin Wortmann, and Andrea Ehrmann.
“Three-dimensional (3D) printing of polymer-metal hybrid materials by fused deposition
modeling”. In: Materials 10.10 (2017). Publisher: Multidisciplinary Digital Publishing Institute,
p. 1199.doi: 10/gckxcb.
[42] Jing Fan, Shuaijun Li, Ziqian Wu, and Zi Chen. “Diffusion and mixing in microfluidic devices”. In:
Microfluidics for Pharmaceutical Applications . Elsevier, 2019, pp. 79–100.
[43] A. Folch, A. Ayon, O. Hurtado, M. A. Schmidt, and M. Toner. “Molding of deep
polydimethylsiloxane microstructures for microfluidics and biological applications”. In: Journal of
Biomechanical Engineering 121.1 (Feb. 1, 1999). Publisher: American Society of Mechanical
Engineers Digital Collection, pp. 28–34.issn: 15288951.doi: 10/c8dqfp.
[44] Flavia Fontana, Mohammad-Ali Shahbazi, Dongfei Liu, Hongbo Zhang, Ermei Mäkilä,
Jarno Salonen, Jouni T Hirvonen, and Hélder A Santos. “Multistaged nanovaccines based on
porous silicon@ acetalated dextran@ cancer cell membrane for cancer immunotherapy”. In:
Advanced Materials 29.7 (2017). Publisher: Wiley Online Library, p. 1603239.
[45] Teruo Fujii. “PDMS-based microfluidic devices for biomedical applications”. In: Microelectronic
Engineering. Vol. 61-62. ISSN: 01679317. Elsevier, July 1, 2002, pp. 907–914.doi: 10/bcz9pz.
[46] Gabriel Gaal, Melissa Mendes, Tiago P de Almeida, Maria HO Piazzetta, Ângelo L Gobbi,
Antonio Riul Jr, and Varlei Rodrigues. “Simplified fabrication of integrated microfluidic devices
using fused deposition modeling 3D printing”. In: Sensors and Actuators B: Chemical 242 (2017).
Publisher: Elsevier, pp. 35–40.doi: 10.1016/j.snb.2016.10.110.
[47] Marie Gaumet, Angelica Vargas, Robert Gurny, and Florence Delie. “Nanoparticles for drug
delivery: the need for precision in reporting particle size parameters”. In: European journal of
pharmaceutics and biopharmaceutics 69.1 (2008). Publisher: Elsevier, pp. 1–9.doi: 10/bq37v6.
[48] Elif Gencturk, Senol Mutlu, and Kutlu O. Ulgen. “Advances in microfluidic devices made from
thermoplastics used in cell biology and analyses”. In: Biomicrofluidics 11.5 (Sept. 1, 2017).
Publisher: American Institute of Physics Inc., p. 51502.issn: 19321058.doi: 10/ggh42w.
[49] George Georgiou, Christos Stathopoulos, Patrick S Daugherty, Amiya R Nayak, Brent L Iverson,
and Roy Curtiss III. “Display of heterologous proteins on the surface of microorganisms: from the
screening of combinatorial libraries to live recombinant vaccines”. In: Nature biotechnology 15.1
(1997). Publisher: Springer, pp. 29–34.doi: 10/dk85kw.
89
[50] Julien Giboz, Thierry Copponnex, and Patrice Mélé. “Microinjection molding of thermoplastic
polymers: A review”. In: Journal of Micromechanics and Microengineering 17.6 (June 1, 2007).
Publisher: IOP Publishing, R96.issn: 09601317.doi: 10/d67wpx.
[51] Hua Gong, Michael Beauchamp, Steven Perry, Adam T. Woolley, and Gregory P. Nordin. “Optical
approach to resin formulation for 3D printed microfluidics”. In: RSC Advances 5.129 (2015),
pp. 106621–106632.issn: 20462069.doi: 10.1039/c5ra23855b.
[52] Hua Gong, Bryce P. Bickham, Adam T. Woolley, and Gregory P. Nordin. “Custom 3D printer and
resin for 18µ m× 20µ m microfluidic flow channels”. In: Lab on a Chip 17.17 (2017). Publisher:
Royal Society of Chemistry, pp. 2899–2909.issn: 14730189.doi: 10/cbtf.
[53] Gregory Gregoriadis, PD Leathwood, and Brenda E Ryman. “Enzyme entrapment in liposomes”.
In: FEBS letters 14.2 (1971). Publisher: Wiley Online Library, pp. 95–99.doi: 10/cjdsgj.
[54] Bethany C. Gross, Jayda L. Erkal, Sarah Y. Lockwood, Chengpeng Chen, and Dana M. Spence.
“Evaluation of 3D printing and its potential impact on biotechnology and the chemical sciences”.
In: Analytical Chemistry 86.7 (2014), pp. 3240–3253.issn: 15206882.doi: 10/gctn85.
[55] Garrett G Gross, Jason A Junge, Rudy J Mora, Hyung-Bae Kwon, C Anders Olson,
Terry T Takahashi, Emily R Liman, Graham CR Ellis-Davies, Aaron W McGee,
Bernardo L Sabatini, et al. “Recombinant probes for visualizing endogenous synaptic proteins in
living neurons”. In: Neuron 78.6 (2013). Publisher: Elsevier, pp. 971–985.doi: 10/f45f67.
[56] Oguzhan Gunduz, Zeeshan Ahmad, Eleanor Stride, and Mohan Edirisinghe. “Continuous
generation of ethyl cellulose drug delivery nanocarriers from microbubbles”. In: Pharmaceutical
research 30.1 (2013). Publisher: Springer, pp. 225–237.doi: 10/f4h5b2.
[57] Arcan Güven, Mayreli Ortiz, Magdalena Constanti, and Ciara K O’Sullivan. “Rapid and efficient
method for the size separation of homogeneous fluorescein-encapsulating liposomes”. In: Journal
of liposome research 19.2 (2009). Publisher: Taylor & Francis, pp. 148–154.doi: 10/c6x6ft.
[58] D Jed Harrison, Andreas Manz, Zhonghui Fan, Hans Luedi, and H Michael Widmer. “Capillary
electrophoresis and sample injection systems integrated on a planar glass chip”. In: Analytical
chemistry 64.17 (1992). Publisher: ACS Publications, pp. 1926–1932.doi: 10/fw5ttg.
[59] Kimberly J. Hassett, Kerry E. Benenato, Eric Jacquinet, Aisha Lee, Angela Woods, Olga Yuzhakov,
Sunny Himansu, Jessica Deterling, Benjamin M. Geilich, Tatiana Ketova, Cosmin Mihai,
Andy Lynn, Iain McFadyen, Melissa J. Moore, Joseph J. Senn, Matthew G. Stanton,
Örn Almarsson, Giuseppe Ciaramella, and Luis A. Brito. “Optimization of Lipid Nanoparticles for
Intramuscular Administration of mRNA Vaccines”. In: Molecular Therapy - Nucleic Acids 15 (Apr.
2019), pp. 1–11.issn: 21622531.doi: 10.1016/j.omtn.2019.01.013. (Visited on 07/11/2022).
[60] Yong He, Yan Wu, Jian Zhong Fu, Qing Gao, and Jing Jiang Qiu. “Developments of 3D Printing
Microfluidics and Applications in Chemistry and Biology: a Review”. In: Electroanalysis 28.8
(Aug. 1, 2016). Publisher: Wiley-VCH Verlag, pp. 1658–1678.issn: 15214109.doi:
10.1002/elan.201600043.
90
[61] Franz X. Heinz and Karin Stiasny. “Distinguishing features of current COVID-19 vaccines:
knowns and unknowns of antigen presentation and modes of action”. In: npj Vaccines 6.1 (Dec.
2021), p. 104.issn: 2059-0105.doi: 10/gm7bsj. (Visited on 07/10/2022).
[62] Elizabeth E. Hills, Michael H. Abraham, Anne Hersey, and Chris D. Bevan. “Diffusion coefficients
in ethanol and in water at 298K: Linear free energy relationships”. In: Fluid Phase Equilibria 303.1
(Apr. 15, 2011), pp. 45–55.issn: 0378-3812.doi: 10/d56752. (Visited on 07/13/2022).
[63] Chee Meng Benjamin Ho, Sum Huan Ng, King Ho Holden Li, and Yong Jin Yoon. “3D printed
microfluidics for biological applications”. In: Lab on a Chip 15.18 (2015). Publisher: Royal Society
of Chemistry, pp. 3627–3637.issn: 14730189.doi: 10/gctr35.
[64] Jessica R Hoffman, Ennio Tasciotti, and Roberto Molinaro. “Microfluidic assembly of liposomes
with tunable size and coloading capabilities”. In: Multiple Myeloma. Springer, 2018, pp. 205–214.
[65] Renee R. Hood and Don L. DeVoe. “High-Throughput Continuous Flow Production of Nanoscale
Liposomes by Microfluidic Vertical Flow Focusing”. In: Small 11.43 (Nov. 2015), pp. 5790–5799.
issn: 16136810.doi: 10/f3gcxf. (Visited on 07/15/2022).
[66] Renee R. Hood, Don L. DeVoe, Javier Atencia, Wyatt N. Vreeland, and Donna M. Omiatek. “A
facile route to the synthesis of monodisperse nanoscale liposomes using 3D microfluidic
hydrodynamic focusing in a concentric capillary array”. In: Lab Chip 14.14 (2014), pp. 2403–2409.
issn: 1473-0197, 1473-0189.doi: 10/gm4b9f. (Visited on 07/14/2022).
[67] Xucheng Hou, Tal Zaks, Robert Langer, and Yizhou Dong. “Lipid nanoparticles for mRNA
delivery”. In: Nature Reviews Materials 6.12 (Dec. 2021), pp. 1078–1094.issn: 2058-8437.doi:
10/gmjsn5. (Visited on 07/11/2022).
[68] Charles W Hull. “Apparatus for production of three-dimensional objects by stereolithography”.
In: United States Patent, Appl., No. 638905, Filed (1984).
[69] Chuck Hull. “On Stereolithography”. In: Virtual and Physical Prototyping 7.3 (Sept. 2012).
Publisher: Taylor & Francis Group, p. 177.issn: 17452759.doi: 10.1080/17452759.2012.723409.
[70] Lien Yu Hung, Chih Hung Wang, Chien Yu Fu, Priya Gopinathan, and Bin Gwo Lee.
“Microfluidics in the selection of affinity reagents for the detection of cancer: Paving a way
towards future diagnostics”. In: Lab on a Chip 16.15 (2016). Publisher: Royal Society of Chemistry,
pp. 2759–2774.issn: 14730189.doi: 10/gqjqcs.
[71] Richard J Jackson and Tim Hunt. “[4] Preparation and use of nuclease-treated rabbit reticulocyte
lysates for the translation of eukaryotic messenger RNA”. In: Methods in enzymology. Vol. 96.
Elsevier, 1983, pp. 50–74.
[72] Richard J. Jackson and Tim Hunt. “Preparation and use of nuclease-treated rabbit reticulocyte
lysates for the translation of eukaryotic messenger RNA”. In: Methods in Enzymology 96 (C Jan. 1,
1983). Publisher: Methods Enzymol, pp. 50–74.issn: 15577988.doi: 10/d7gbjt.
91
[73] Stephen C Jacobson, Roland Hergenroder, Lance B Koutny, and J Michael Ramsey. “High-speed
separations on a microchip”. In: Analytical Chemistry 66.7 (1994). Publisher: ACS Publications,
pp. 1114–1118.doi: 10/fbfcrw.
[74] Andreas Jahn, Wyatt N Vreeland, Don L DeVoe, Laurie E Locascio, and Michael Gaitan.
“Microfluidic directed formation of liposomes of controlled size”. In: Langmuir 23.11 (2007).
Publisher: ACS Publications, pp. 6289–6293.doi: 10/drf5tn.
[75] Farzad Jalali-Yazdi, Jasmine M Corbin, Terry T Takahashi, and Richard W Roberts. “Robust,
Quantitative Analysis of Proteins using Peptide Immunoreagents, in Vitro Translation, and an
Ultrasensitive Acoustic Resonant Sensor”. In: Analytical chemistry 86.10 (2014). Publisher: ACS
Publications, pp. 4715–4722.doi: 10/f547hc.
[76] Farzad Jalali-Yazdi, Lan Huong Lai, Terry T Takahashi, and Richard W Roberts.
“High-Throughput Measurement of Binding Kinetics by mRNA Display and Next-Generation
Sequencing”. In: Angewandte Chemie International Edition 55.12 (2016). Publisher: Wiley Online
Library, pp. 4007–4010.doi: 10/f3k5k7.
[77] Farzad Jalali-Yazdi, Terry T Takahashi, and Richard W Roberts. “General, Label-Free Method for
Determining K d and Ligand Concentration Simultaneously”. In: Analytical chemistry 87.23
(2015). Publisher: ACS Publications, pp. 11755–11762.doi: 10/f73vmm.
[78] Jessie S Jeon, Seok Chung, Roger D Kamm, and Joseph L Charest. “Hot embossing for fabrication
of a microfluidic 3D cell culture platform”. In: (0). doi: 10/ccp989.
[79] Dae-Eun Jeong, Matthew McCoy, Karen Artiles, Orkan Ilbay, Andrew Fire, Kari Nadeau,
Helen Park, Brooke Betts, Scott Boyd, Ramona Hoh, and Massa Shoura. “Assemblies of putative
SARS-CoV2-spike-encoding mRNA sequences for vaccines BNT- 162b2 and mRNA-1273”. In:
(Apr. 14, 2021).url: https://raw.githubusercontent.com/NAalytics/Assemblies-of-putative-SARS-
CoV2-spike-encoding-mRNA-sequences-for-vaccines-BNT-162b2-and-mRNA-
1273/main/Assemblies%20of%20putative%20SARS-CoV2-spike-
encoding%20mRNA%20sequences%20for%20vaccines%20BNT-162b2%20and%20mRNA-1273.docx.pdf.
[80] Bryan Kaehr and Jason B Shear. “Mask-directed multiphoton lithography”. In: Journal of the
American Chemical Society 129.7 (2007). Publisher: ACS Publications, pp. 1904–1905.doi:
10/cf89v4.
[81] Bryan Kaehr and Jason B Shear. “Multiphoton fabrication of chemically responsive protein
hydrogels for microactuation”. In: Proceedings of the National Academy of Sciences 105.26 (2008).
Publisher: National Acad Sciences, pp. 8850–8854.doi: 10/fw5c8k.
[82] Nauman Khalid, Sania Arif, Isao Kobayashi, and Mitsutoshi Nakajima. “Lab-on-a-chip techniques
for high-throughput proteomics and drug discovery”. In: Microfluidics for Pharmaceutical
Applications. Elsevier, 2019, pp. 371–422.
[83] Kangil Kim, Sin Wook Park, and Sang Sik Yang. “The optimization of PDMS-PMMA bonding
process using silane primer”. In: BioChip Journal 4.2 (June 2010), pp. 148–154.issn: 1976-0280,
2092-7843.doi: 10.1007/s13206-010-4210-0. (Visited on 07/11/2022).
92
[84] David J. Kinahan, Faith Mangwanya, Robert Garvey, Danielle W.Y. Chung, Artur Lipinski,
Lourdes A.N. Julius, Damien King, Mehdi Mohammadi, Rohit Mishra, May Al-Ofi,
Celina Miyazaki, and Jens Ducrée. “Automation of Silica Bead-based Nucleic Acid Extraction on a
Centrifugal Lab-on-a-Disc Platform”. In: Journal of Physics: Conference Series 757.1 (2016).issn:
17426596.doi: 10/gqjqct.
[85] Rimantas Kodzius, Kang Xiao, Jinbo Wu, Xin Yi, Xiuqing Gong, Ian G. Foulds, and Weijia Wen.
“Inhibitory effect of common microfluidic materials on PCR outcome”. In: Sensors and Actuators,
B: Chemical 161.1 (2012). Publisher: Elsevier B.V., pp. 349–358.issn: 09254005.doi:
10.1016/j.snb.2011.10.044.
[86] Georges Köhler and Cesar Milstein. “Continuous cultures of fused cells secreting antibody of
predefined specificity”. In: nature 256.5517 (1975). Publisher: Springer, pp. 495–497.doi:
10.1038/256495a0.
[87] Carol A. Kreader. “Relief of amplification inhibition in PCR with bovine serum albumin or T4
gene 32 protein”. In: Applied and Environmental Microbiology 62.3 (1996). Publisher: American
Society for Microbiology, pp. 1102–1106.issn: 00992240.doi: 10/gkcq53.
[88] Christopher N LaFratta, John T Fourkas, Tommaso Baldacchini, and Richard A Farrer.
“Multiphoton fabrication”. In: Angewandte Chemie International Edition 46.33 (2007). Publisher:
Wiley Online Library, pp. 6238–6258.doi: 10/c5fhxd.
[89] Bin Gwo Lee, Shu Hui Chen, Guan Ruey Huang, Wang Chou Sung, and Yen Heng Lin.
“Microfabricated plastic chips by hot embossing methods and their applications for DNA
separation and detection”. In: Sensors and Actuators, B: Chemical 75.1 (Apr. 30, 2001). Publisher:
Elsevier, pp. 142–148.issn: 09254005.doi: 10/ckqc5q.
[90] Wonjae Lee, Donghoon Kwon, Woong Choi, Gyoo Yeol Jung, Anthony K. Au, Albert Folch, and
Sangmin Jeon. “3D-Printed Microfluidic Device for the Detection of Pathogenic Bacteria Using
Size-based Separation in Helical Channel with Trapezoid Cross-Section”. In: Scientific Reports 5.1
(July 15, 2015), p. 7717.issn: 2045-2322.doi: 10/gnthqp. (Visited on 07/13/2022).
[91] Alex K. K. Leung, Yuen Yi C. Tam, Sam Chen, Ismail M. Hafez, and Pieter R. Cullis. “Microfluidic
Mixing: A General Method for Encapsulating Macromolecules in Lipid Nanoparticle Systems”. In:
The Journal of Physical Chemistry B 119.28 (July 16, 2015), pp. 8698–8706.issn: 1520-6106,
1520-5207.doi: 10/ghxht5. (Visited on 07/11/2022).
[92] Hsiang-I Liao, C Anders Olson, Seungmin Hwang, Hongyu Deng, Elaine Wong, Ralph S Baric,
Richard W Roberts, and Ren Sun. “mRNA display design of fibronectin-based intrabodies that
detect and inhibit severe acute respiratory syndrome coronavirus nucleocapsid protein”. In:
Journal of Biological Chemistry 284.26 (2009). Publisher: ASBMB, pp. 17512–17520.doi: 10/bc54bc.
[93] Jong-Min Lim, Archana Swami, Laura M Gilson, Sunandini Chopra, Sungyoung Choi, Jun Wu,
Robert Langer, Rohit Karnik, and Omid C Farokhzad. “Ultra-high throughput synthesis of
nanoparticles with homogeneous size distribution using a coaxial turbulent jet mixer”. In: ACS
nano 8.6 (2014). Publisher: ACS Publications, pp. 6056–6065.doi: 10.1021/nn501371n.
93
[94] Tung Yi Lin, Truong Do, Patrick Kwon, and Peter B. Lillehoj. “3D printed metal molds for hot
embossing plastic microfluidic devices”. In: Lab on a Chip 17.2 (Jan. 21, 2017). Publisher: Royal
Society of Chemistry, pp. 241–247.issn: 14730189.doi: 10/gqjqcv.
[95] Wan-Zhen Sophie Lin, William E. Evenson, W. Kristian Vu Bostic, Richard W. Roberts, and
Noah Malmstadt. “Compatibility of Popular Three-Dimensional Printed Microfluidics Materials
with In-Vitro Enzymatic Reactions”. In: ACS Applied Bio Materials 5.2 (Feb. 21, 2022), pp. 818–824.
issn: 2576-6422, 2576-6422.doi: 10/gqjqcw. (Visited on 07/23/2022).
[96] Wan-Zhen Sophie Lin and Noah Malmstadt. “Liposome production and concurrent loading of
drug simulants by microfluidic hydrodynamic focusing”. In: European Biophysics Journal 48.6
(Sept. 2019), pp. 549–558.issn: 0175-7571, 1432-1017.doi: 10/gqjqcx. (Visited on 07/12/2022).
[97] David C Litzinger, Antoinette MJ Buiting, Nico van Rooijen, and Leaf Huang. “Effect of liposome
size on the circulation time and intraorgan distribution of amphipathic poly (ethylene
glycol)-containing liposomes”. In: Biochimica et Biophysica Acta (BBA)-Biomembranes 1190.1
(1994). Publisher: Elsevier, pp. 99–107.doi: 10/fppbft.
[98] Der-Zen Liu, Wen-Yih Chen, Li-Min Tasi, and Shu-Ping Yang. “Microcalorimetric and shear
studies on the effects of cholesterol on the physical stability of lipid vesicles”. In: Colloids and
Surfaces A: Physicochemical and Engineering Aspects 172.1 (2000). Publisher: Elsevier, pp. 57–67.
doi: 10/dtmm2m.
[99] Rihe Liu, Jeffrey E Barrick, Jack W Szostak, and Richard W Roberts. “[19] Optimized synthesis of
RNA-protein fusions for in vitro protein selection”. In: (2000). Publisher: Elsevier.doi:
10.1016/s0076-6879(00)18058-9.
[100] Yanli Liu, Jonathan D Adams, Kelisha Turner, Frank V Cochran, Sanjiv Sam Gambhir, and
H Tom Soh. “Controlling the selection stringency of phage display using a microfluidic device”.
In: Lab on a chip 9.8 (2009). Publisher: Royal Society of Chemistry, pp. 1033–1036.doi:
10.1039/b820985e.
[101] Xinhui Lou, Jiangrong Qian, Yi Xiao, Lisan Viel, Aren E Gerdon, Eric T Lagally, Paul Atzberger,
Theodore M Tarasow, Alan J Heeger, and H Tom Soh. “Micromagnetic selection of aptamers in
microfluidic channels”. In: Proceedings of the National Academy of Sciences 106.9 (2009). Publisher:
National Acad Sciences, pp. 2989–2994.doi: 10/bv4s7c.
[102] Masatoshi Maeki, Yuka Fujishima, Yusuke Sato, Takao Yasui, Noritada Kaji, Akihiko Ishida,
Hirofumi Tani, Yoshinobu Baba, Hideyoshi Harashima, and Manabu Tokeshi. “Understanding the
formation mechanism of lipid nanoparticles in microfluidic devices with chaotic micromixers”. In:
PLoS ONE 12.11 (Nov. 28, 2017), e0187962.issn: 1932-6203.doi: 10/gcmpbp. (Visited on 07/12/2022).
[103] Masatoshi Maeki, Niko Kimura, Yusuke Sato, Hideyoshi Harashima, and Manabu Tokeshi.
“Advances in microfluidics for lipid nanoparticles and extracellular vesicles and applications in
drug delivery systems”. In: Advanced drug delivery reviews 128 (2018). Publisher: Elsevier,
pp. 84–100.doi: 10/gdx86f.
94
[104] Andréas Manz, N Graber, and H áM Widmer. “Miniaturized total chemical analysis systems: a
novel concept for chemical sensing”. In: Sensors and actuators B: Chemical 1.1 (1990). Publisher:
Elsevier, pp. 244–248.doi: 10/bk7fcm.
[105] Mao Mao, Jiankang He, Xiao Li, Bing Zhang, Qi Lei, Yaxiong Liu, and Dichen Li. “The emerging
frontiers and applications of high-resolution 3D printing”. In:Micromachines 8.4 (2017). Publisher:
Multidisciplinary Digital Publishing Institute, p. 113.doi: 10/ggs4nr.
[106] Ulrich Massing and Stefan Fuxius. “Liposomal formulations of anticancer drugs: selectivity and
effectiveness”. In: Drug resistance updates 3.3 (2000). Publisher: Elsevier, pp. 171–177.doi:
10/c73khv.
[107] Maria Matschuk and Niels B. Larsen. “Injection molding of high aspect ratio sub-100 nm
nanostructures”. In: Journal of Micromechanics and Microengineering 23.2 (Feb. 21, 2013).
Publisher: IOP Publishing, p. 025003.issn: 09601317.doi: 10/gqjqcz.
[108] Giuseppe Maulucci, Marco De Spirito, Giuseppe Arcovito, Federico Boffi,
Agostina Congiu Castellano, and Giuseppe Briganti. “Particle size distribution in DMPC vesicles
solutions undergoing different sonication times”. In: Biophysical journal 88.5 (2005). Publisher:
Elsevier, pp. 3545–3550.doi: 10.1529/biophysj.104.048876.
[109] J Cooper McDonald, Michael L Chabinyc, Steven J Metallo, Janelle R Anderson,
Abraham D Stroock, and George M Whitesides. “Prototyping of microfluidic devices in poly
(dimethylsiloxane) using solid-object printing”. In: Analytical chemistry 74.7 (2002). Publisher:
ACS Publications, pp. 1537–1545.doi: 10/bss7z3.
[110] J. Cooper McDonald, Steven J. Metallo, and George M. Whitesides. “Fabrication of a Configurable,
Single-Use Microfluidic Device”. In: Analytical Chemistry 73.23 (Dec. 1, 2001), pp. 5645–5650.
issn: 0003-2700, 1520-6882.doi: 10/dkcfhr. (Visited on 07/11/2022).
[111] Roberto Mendez and Santanu Banerjee. “Sonication-Based Basic Protocol for Liposome
Synthesis”. In: Lipidomics: Methods and Protocols. Ed. by Sanjoy K. Bhattacharya. Methods in
Molecular Biology. New York, NY: Springer, 2017, pp. 255–260.isbn: 978-1-4939-6996-8.doi:
10.1007/978-1-4939-6996-8_21. (Visited on 07/12/2022).
[112] Michael Montero, Shad Roundy, Dan Odell, Sung-Hoon Ahn, and Paul K Wright. “Material
characterization of fused deposition modeling (FDM) ABS by designed experiments”. In: Society
of Manufacturing Engineers 10.13552540210441166 (2001), pp. 1–21.
[113] Rudy J Mora, Richard W Roberts, and Don B Arnold. “Recombinant probes reveal dynamic
localization of CaMKIIα within somata of cortical neurons”. In: Journal of Neuroscience 33.36
(2013). Publisher: Soc Neuroscience, pp. 14579–14590.doi: 10/f49q3w.
[114] Rajendrani Mukhopadhyay. When PDMS isn’t the best. 2007.
[115] Geoffrey Mulberry, Kevin A. White, Manjusha Vaidya, Kiminobu Sugaya, and Brian N. Kim. “3D
printing and milling a real-time PCR device for infectious disease diagnostics”. In: PLoS ONE 12.6
(2017). ISBN: 1111111111, pp. 1–18.issn: 19326203.doi: 10.1371/journal.pone.0179133.
95
[116] A Nagayasu, K Uchiyama, and H Kiwada. “The size of liposomes: a factor which affects their
targeting efficiency to tumors and therapeutic activity of liposomal antitumor drugs”. In:
Advanced Drug Delivery Reviews 40.1 (Nov. 1999), pp. 75–87.issn: 0169409X.doi: 10/crq96z.
(Visited on 07/17/2022).
[117] Zuzana Nejedlá, David Poustka, Regina Herma, Michaela Liegertová, Marcel Štofik, Jiří Smejkal,
Václav Šícha, Pavel Kaule, and Jan Malý. “Class II biocompatible E-Shell 300 3D printing material
causes severe developmental toxicity in Danio rerio embryos and reduced cell proliferation in
vitro – implications for 3D printed microfluidics”. In: RSCAdvances 11.27 (May 4, 2021). Publisher:
Royal Society of Chemistry, pp. 16252–16267.issn: 2046-2069.doi: 10/gqjqc2.
[118] Jennifer Nelson, Elizabeth W. Sorensen, Shrutika Mintri, Amy E. Rabideau, Wei Zheng,
Gilles Besin, Nikhil Khatwani, Stephen V. Su, Edward J. Miracco, William J. Issa, Stephen Hoge,
Matthew G. Stanton, and John L. Joyal. “Impact of mRNA chemistry and manufacturing process
on innate immune activation”. In: Science Advances 6.26 (June 26, 2020), eaaz6893.issn:
2375-2548.doi: 10/gmpm3t. (Visited on 07/11/2022).
[119] Tuan D Ngo, Alireza Kashani, Gabriele Imbalzano, Kate TQ Nguyen, and David Hui. “Additive
manufacturing (3D printing): A review of materials, methods, applications and challenges”. In:
Composites Part B: Engineering 143 (2018). Publisher: Elsevier, pp. 172–196.doi: 10/gdmhrk.
[120] Seung Soo Oh, Kareem M Ahmad, Minseon Cho, Seon Kim, Yi Xiao, and H Tom Soh. “Improving
aptamer selection efficiency through volume dilution, magnetic concentration, and continuous
washing in microfluidic channels”. In: Analytical chemistry 83.17 (2011). Publisher: ACS
Publications, pp. 6883–6889.doi: 10/ft2fmd.
[121] C Anders Olson, Jonathan D Adams, Terry T Takahashi, Hangfei Qi, Shannon M Howell,
Ting-Ting Wu, Richard W Roberts, Ren Sun, and H Tom Soh. “Rapid mRNA-Display Selection of
an IL-6 Inhibitor Using Continuous-Flow Magnetic Separation”. In: Angewandte Chemie
International Edition 50.36 (2011). Publisher: Wiley Online Library, pp. 8295–8298.doi: 10/ccgtkx.
[122] C Anders Olson, Jeff Nie, Jonathan Diep, Ibrahim Al-Shyoukh, Terry T Takahashi,
Laith Q Al-Mawsawi, Jennifer M Bolin, Angela L Elwell, Scott Swanson, Ron Stewart, et al.
“Single-Round, Multiplexed Antibody Mimetic Design through mRNA Display”. In: Angewandte
Chemie International Edition 51.50 (2012). Publisher: Wiley Online Library, pp. 12449–12453.doi:
10/fz9c8v.
[123] A. K. Pantazis, G. Papadakis, K. Parasyris, A. Stavrinidis, and E. Gizeli. “3D-printed bioreactors for
DNA amplification: application to companion diagnostics”. In: Sensors and Actuators, B: Chemical
319 (February 2020). Publisher: Elsevier, p. 128161.issn: 09254005.doi:
10.1016/j.snb.2020.128161.
[124] Hong Key Park, Mikyung Shin, Bongkyun Kim, Jin Woo Park, and Haeshin Lee. “A visible
light-curable yet visible wavelength-transparent resin for stereolithography 3D printing”. In:NPG
Asia Materials 10.4 (2018). ISBN: 4142701800 Publisher: Springer US, pp. 82–89.issn: 18844057.
doi: 10/gqjqc4.
96
[125] Jayanta Kumar Patra, Gitishree Das, Leonardo Fernandes Fraceto,
Estefania Vangelie Ramos Campos, Maria del Pilar Rodriguez-Torres, Laura Susana Acosta-Torres,
Luis Armando Diaz-Torres, Renato Grillo, Mallappa Kumara Swamy, Shivesh Sharma, et al.
“Nano based drug delivery systems: recent developments and future prospects”. In: Journal of
nanobiotechnology 16.1 (2018). Publisher: BioMed Central, p. 71.doi: 10.1186/s12951-018-0392-8.
[126] Dan Peer, Jeffrey M Karp, Seungpyo Hong, Omid C Farokhzad, Rimona Margalit, and
Robert Langer. “Nanocarriers as an emerging platform for cancer therapy”. In: Nature
nanotechnology 2.12 (2007). Publisher: Nature Publishing Group, pp. 751–760.doi: 10/bzf2j5.
[127] Duc Truong Pham and Rosemary S Gault. “A comparison of rapid prototyping technologies”. In:
International Journal of machine tools and manufacture 38.10 (1998). Publisher: Elsevier,
pp. 1257–1287.doi: 10.1016/s0890-6955(97)00137-5.
[128] “Phone call with Precision NanoSystems Sales Department”. In: (July 12, 2022).
[129] Charles Pidgeon, Susan McNeely, Tim Schmidt, and John E Johnson. “Multilayered vesicles
prepared by reverse-phase evaporation: liposome structure and optimum solute entrapment”. In:
Biochemistry 26.1 (1987). Publisher: ACS Publications, pp. 17–29.doi: 10/bmtbqq.
[130] Fernando P. Polack, Stephen J. Thomas, Nicholas Kitchin, Judith Absalon, Alejandra Gurtman,
Stephen Lockhart, John L. Perez, Gonzalo Pérez Marc, Edson D. Moreira, Cristiano Zerbini,
Ruth Bailey, Kena A. Swanson, Satrajit Roychoudhury, Kenneth Koury, Ping Li, Warren V. Kalina,
David Cooper, Robert W. Frenck, Laura L. Hammitt, Özlem Türeci, Haylene Nell, Axel Schaefer,
Serhat Ünal, Dina B. Tresnan, Susan Mather, Philip R. Dormitzer, Uğur Şahin, Kathrin U. Jansen,
and William C. Gruber. “Safety and Efficacy of the BNT162b2 mRNA Covid-19 Vaccine”. In: New
England Journal of Medicine 383.27 (Dec. 31, 2020), pp. 2603–2615.issn: 0028-4793, 1533-4406.
doi: 10/ghn625. (Visited on 07/12/2022).
[131] A. Ranjit Prakash, Matthias Amrein, and Karan V.I.S. Kaler. “Characteristics and impact of Taq
enzyme adsorption on surfaces in microfluidic devices”. In: Microfluidics and Nanofluidics 4.4
(Apr. 2, 2008). Publisher: Springer, pp. 295–305.issn: 16134982.doi: 10/dbsszp.
[132] Jiangrong Qian, Xinhui Lou, Yanting Zhang, Yi Xiao, and H Tom Soh. “Generation of highly
specific aptamers via micromagnetic selection”. In: Analytical chemistry 81.13 (2009). Publisher:
ACS Publications, pp. 5490–5495.doi: 10/ffv7nb.
[133] Kiran Raj M and Suman Chakraborty. “PDMS microfluidics: A mini review”. In: JournalofApplied
Polymer Science 137.27 (July 15, 2020). Publisher: John Wiley and Sons Inc., p. 48958.issn:
10974628.doi: 10/gmfksv.
[134] Kangning Ren, Jianhua Zhou, and Hongkai Wu. “Materials for microfluidic chip fabrication”. In:
Accounts of Chemical Research 46.11 (Nov. 19, 2013). Publisher: American Chemical Society,
pp. 2396–2406.issn: 00014842.doi: 10/f5mc76.
97
[135] Patrick Risch, Frederik Kotz, Dorothea Helmer, and Bastian E. Rapp. “3D printing of highly
fluorinated methacrylates for the fabrication of transparent and chemically-resistant microfluidic
devices”. In: 22nd International Conference on Miniaturized Systems for Chemistry and Life
Sciences, MicroTAS 2018. Vol. 4. ISSN: 16057422. 2018, pp. 2163–2165.isbn: 978-1-5108-9757-1.doi:
10/gqjqc5.
[136] Eric K Sackmann, Anna L Fulton, and David J Beebe. “The present and future role of microfluidics
in biomedical research”. In: Nature 507.7491 (2014). Publisher: Nature Publishing Group,
pp. 181–189.doi: 10.1038/nature13118.
[137] Vishal Sahore, Steven R. Doonan, and Ryan C. Bailey. “Droplet microfluidics in thermoplastics:
device fabrication, droplet generation, and content manipulation using integrated electric and
magnetic fields”. In: Analytical Methods 10.35 (Sept. 21, 2018). Publisher: Royal Society of
Chemistry, pp. 4264–4274.issn: 17599679.doi: 10/gj2r89.
[138] Sharma T Sanjay, Wan Zhou, Maowei Dou, Hamed Tavakoli, Lei Ma, Feng Xu, and XiuJun Li.
“Recent advances of controlled drug delivery using microfluidic platforms”. In: Advanced drug
delivery reviews 128 (2018). Publisher: Elsevier, pp. 3–28.doi: 10/gdzfjb.
[139] Maximilian Schlosshauer and David Baker. “Realistic protein–protein association rates from a
simple diffusional model neglecting long-range interactions, free energy barriers, and landscape
ruggedness”. In: Protein Science 13.6 (2004). Publisher: Wiley Online Library, pp. 1660–1669.doi:
10.1110/ps.03517304.
[140] Linde Schoenmaker, Dominik Witzigmann, Jayesh A. Kulkarni, Rein Verbeke, Gideon Kersten,
Wim Jiskoot, and Daan J.A. Crommelin. “mRNA-lipid nanoparticle COVID-19 vaccines: Structure
and stability”. In: International Journal of Pharmaceutics 601 (May 2021), p. 120586.issn:
03785173.doi: 10.1016/j.ijpharm.2021.120586. (Visited on 07/10/2022).
[141] Carl Schubert, Mark C Van Langeveld, and Larry A Donoso. “Innovations in 3D printing: a 3D
overview from optics to organs”. In: British Journal of Ophthalmology 98.2 (2014). Publisher: BMJ
Publishing Group Ltd, pp. 159–161.doi: 10.1136/bjophthalmol-2013-304446.
[142] Alexandros Selimis, Vladimir Mironov, and Maria Farsari. “Direct laser writing: Principles and
materials for scaffold 3D printing”. In: Microelectronic Engineering 132 (2015). Publisher: Elsevier,
pp. 83–89.doi: 10.1016/j.mee.2014.10.001.
[143] Seyed Ali Mousavi Shaegh, Adel Pourmand, Mahboubeh Nabavinia, Huseyin Avci, Ali Tamayol,
Pooria Mostafalu, Habib Badri Ghavifekr, Esmaeil Najafi Aghdam, Mehmet Remzi Dokmeci,
Ali Khademhosseini, and Yu Shrike Zhang. “Rapid prototyping of whole-thermoplastic
microfluidics with built-in microvalves using laser ablation and thermal fusion bonding”. In:
Sensors and Actuators, B: Chemical 255 (Feb. 1, 2018). Publisher: Elsevier B.V., pp. 100–109.issn:
09254005.doi: 10/ghm33x.
[144] P. P. Shiu, G. K. Knopf, M. Ostojic, and S. Nikumb. “Rapid fabrication of tooling for microfluidic
devices via laser micromachining and hot embossing”. In: Journal of Micromechanics and
Microengineering 18.2 (Feb. 1, 2008). Publisher: IOP Publishing, p. 025012.issn: 09601317.doi:
10/bcq527.
98
[145] Samuel K Sia and George M Whitesides. “Microfluidic devices fabricated in poly
(dimethylsiloxane) for biological studies”. In: Electrophoresis 24.21 (2003). Publisher: Wiley Online
Library, pp. 3563–3576.doi: 10/cpm5qg.
[146] Madhusudan Singh, Hanna M Haverinen, Parul Dhagat, and Ghassan E Jabbour. “Inkjet
printing—process and its applications”. In: Advanced materials 22.6 (2010). Publisher: Wiley
Online Library, pp. 673–685.
[147] Rajesh Singh and James W. Lillard. “Nanoparticle-based targeted drug delivery”. In: Experimental
and Molecular Pathology 86.3 (June 2009), pp. 215–223.issn: 1096-0945.doi: 10/d67bsd.
[148] Anirban Sinha, Priya Gopinathan, Da Yi Chung, Hsin Ying Lin, Kuang Hsien Li, Hsi Pin Ma,
Po Chiun Huang, Shu Chu Shiesh, and Bin Gwo Lee. “An integrated microfluidic platform to
perform uninterrupted SELEX cycles to screen affinity reagents specific to cardiovascular
biomarkers”. In: Biosensors and Bioelectronics 122 (September 2018), pp. 104–112.issn: 18734235.
doi: 10/gqjqc6.
[149] Elaheh Sirjani, Maciej Migas, Peter J. Cragg, and Marcus K. Dymond. “3D printed UV/VIS
detection systems constructed from transparent filaments and immobilised enzymes”. In: Additive
Manufacturing 33 (January 2020). Publisher: Elsevier, p. 101094.issn: 22148604.doi: 10/gqjqc7.
[150] George P Smith and Valery A Petrenko. “Phage display”. In: Chemical reviews 97.2 (1997).
Publisher: ACS Publications, pp. 391–410.doi: 10/c8z5w6.
[151] Ryan D. Sochol, Eric Sweet, Casey C. Glick, Sung Yueh Wu, Chen Yang, Michael Restaino, and
Liwei Lin. “3D printed microfluidics and microelectronics”. In: Microelectronic Engineering 189
(2018), pp. 52–68.issn: 01679317.doi: 10/gfb635.
[152] F William Studier. “Protein production by auto-induction in high-density shaking cultures”. In:
Protein expression and purification 41.1 (2005). Publisher: Elsevier, pp. 207–234.doi:
10.1016/j.pep.2005.01.016.
[153] Francis Szoka and Demetrios Papahadjopoulos. “Procedure for preparation of liposomes with
large internal aqueous space and high capture by reverse-phase evaporation”. In: Proceedings of
the national academy of sciences 75.9 (1978). Publisher: National Acad Sciences, pp. 4194–4198.
doi: 10/df5pkx.
[154] S. Takenaga, B. Schneider, E. Erbay, M. Biselli, Th Schnitzler, M. J. Schöning, and T. Wagner.
“Fabrication of biocompatible lab-on-chip devices for biomedical applications by means of a
3D-printing process”. In: Physica Status Solidi (A) Applications and Materials Science 212.6 (June 1,
2015). Publisher: Wiley-VCH Verlag, pp. 1347–1352.issn: 18626319.doi: 10.1002/pssa.201532053.
[155] Yung-Chieh Tan, Kanaka Hettiarachchi, Maria Siu, Yen-Ru Pan, and Abraham Phillip Lee.
“Controlled microfluidic encapsulation of cells, proteins, and microbeads in lipid vesicles”. In:
Journal of the American Chemical Society 128.17 (2006). Publisher: ACS Publications,
pp. 5656–5658.doi: 10/bhhm4g.
99
[156] Theresa B. Taylor, Emily S. Winn-Deen, Enrico Picozza, Timothy M. Woudenberg, and
Michael Albin. “Optimization of the performance of the polymerase chain reaction in
silicon-based microstructures”. In: Nucleic Acids Research 25.15 (Aug. 1, 1997). Publisher: Oxford
University Press, pp. 3164–3168.issn: 03051048.doi: 10/cpzwgj.
[157] Rumiana Tenchov, Robert Bird, Allison E. Curtze, and Qiongqiong Zhou. “Lipid Nanoparticles -
From Liposomes to mRNA Vaccine Delivery, a Landscape of Research Diversity and
Advancement”. In: ACS Nano 15.11 (Nov. 23, 2021), pp. 16982–17015.issn: 1936-0851.doi:
10/gk4m5m. (Visited on 07/11/2022).
[158] Stephen C Terry, John H Jerman, and James B Angell. “A gas chromatographic air analyzer
fabricated on a silicon wafer”. In: IEEE transactions on electron devices 26.12 (1979). Publisher:
IEEE, pp. 1880–1886.doi: 10/dfq79m.
[159] Vladimir P Torchilin. “Recent advances with liposomes as pharmaceutical carriers”. In: Nature
reviews Drug discovery 4.2 (2005). Publisher: Nature Publishing Group, pp. 145–160.doi:
10.1038/nrd1632.
[160] A. M. Tothill, M. Partridge, S. W. James, and R. P. Tatam. “Fabrication and optimisation of a fused
filament 3D-printed microfluidic platform”. In: Journal of Micromechanics and Microengineering
27.3 (2017). Publisher: IOP Publishing, p. 035018.issn: 13616439.doi: 10/gjrbvn.
[161] Charalampos Tzivelekis, Pavlos Sgardelis, Kevin Waldron, Richard Whalley, Dehong Huo, and
Kenny Dalgarno. “Fabrication routes via projection stereolithography for 3D-printing of
microfluidic geometries for nucleic acid amplification”. In: PLOS ONE 15.10 (Oct. 28, 2020). Ed. by
Andreas Offenhausser, e0240237. issn: 1932-6203.doi: 10/gqjqc8. (Visited on 07/24/2022).
[162] Precision NanoSystems ULC. “NanoAssemblr Ignite and Ignite +, Power to Advance”. In: (2022).
issn: naignite-BR-0622.url: https://www.precisionnanosystems.com/platform-
technologies/product-comparison/ignite#Downloadbrochure.
[163] Arturo Urrios, Cesar Parra-Cabrera, Nirveek Bhattacharjee, Alan M. Gonzalez-Suarez,
Luis G. Rigat-Brugarolas, Umashree Nallapatti, Josep Samitier, Cole A. Deforest, Francesc Posas,
José L. Garcia-Cordero, and Albert Folch. “3D-printing of transparent bio-microfluidic devices in
PEG-DA”. In: Lab on a Chip 16.12 (2016). Publisher: Royal Society of Chemistry, pp. 2287–2294.
issn: 14730189.doi: 10.1039/c6lc00153j.
[164] Anand Venkataraman, Kun Yang, Jose Irizarry, Mark Mackiewicz, Paolo Mita, Zheng Kuang,
Lin Xue, Devlina Ghosh, Shuang Liu, Pedro Ramos, et al. “A toolbox of
immunoprecipitation-grade monoclonal antibodies to human transcription factors”. In: Nature
methods 15.5 (2018). Publisher: Nature Publishing Group, p. 330.doi: 10/gqjqdh.
[165] C. Lee Ventola. “Medical Applications for 3D Printing: Current and Projected Uses”. In: P & T: A
Peer-Reviewed Journal for Formulary Management 39.10 (Oct. 2014), pp. 704–711.issn: 1052-1372.
[166] Sidra Waheed, Joan M. Cabot, Niall P. Macdonald, Trevor Lewis, Rosanne M. Guijt, Brett Paull,
and Michael C. Breadmore. “3D printed microfluidic devices: Enablers and barriers”. In: Lab on a
Chip 16.11 (2016). Publisher: Royal Society of Chemistry, pp. 1993–2013.issn: 14730189.doi:
10/gfb75v.
100
[167] Lu Wang, Lanja R. Karadaghi, Richard L. Brutchey, and Noah Malmstadt. “Self-optimizing parallel
millifluidic reactor for scaling nanoparticle synthesis”. In: Chemical Communications 56.26 (2020),
pp. 3745–3748.issn: 1359-7345, 1364-548X.doi: 10/gqjqc9. (Visited on 07/14/2022).
[168] George M Whitesides. “The origins and the future of microfluidics”. In: Nature 442.7101 (2006).
Publisher: Nature Publishing Group, pp. 368–373.doi: 10/bwq2nd.
[169] Peter Wilding, Mann A. Shoffner, and Larry J. Kricka. “PCR in a silicon microstructure”. In:
Clinical Chemistry. Vol. 40. ISSN: 00099147 Issue: 9. American Association for Clinical Chemistry
Inc., Sept. 1, 1994, pp. 1815–1818.doi: 10.1093/clinchem/40.9.1815.
[170] Younan Xia and George M Whitesides. “Soft lithography”. In: Annual review of materials science
28.1 (1998). Publisher: Annual Reviews 4139 El Camino Way, PO Box 10139, Palo Alto, CA
94303-0139, USA, pp. 153–184.doi: 10/crv4nx.
[171] Liang Xiao, Kuo-Chan Hung, Terry T Takahashi, Kye-Il Joo, Matthew Lim, Richard W Roberts,
and Pin Wang. “Antibody-mimetic ligand selected by mRNA display targets DC-SIGN for
dendritic cell-directed antigen delivery”. In: ACS chemical biology 8.5 (2013). Publisher: ACS
Publications, pp. 967–977.doi: 10.1021/cb300680c.
[172] Bin-Bin Xu, Ran Zhang, Xue-Qing Liu, Huan Wang, Yong-Lai Zhang, Hao-Bo Jiang, Lei Wang,
Zhuo-Chen Ma, Jin-Feng Ku, Feng-Shou Xiao, et al. “On-chip fabrication of silver microflower
arrays as a catalytic microreactor for allowing in situ SERS monitoring”. In: Chemical
Communications 48.11 (2012). Publisher: Royal Society of Chemistry, pp. 1680–1682.doi:
10/fnvc7n.
[173] Yang Xu, Fangjie Qi, Huachao Mao, Songwei Li, Yizhen Zhu, Jingwen Gong, Lu Wang,
Noah Malmstadt, and Yong Chen. “In-situ transfer vat photopolymerization for transparent
microfluidic device fabrication”. In: Nature Communications 13.1 (Dec. 2022), p. 918.issn:
2041-1723.doi: 10/gqjqdc. (Visited on 07/24/2022).
[174] Zheqi Xu, Changhai Lu, Jason Riordon, David Sinton, and Matthew G Moffitt. “Microfluidic
manufacturing of polymeric nanoparticles: comparing flow control of multiscale structure in
single-phase staggered herringbone and two-phase reactors”. In: Langmuir 32.48 (2016).
Publisher: ACS Publications, pp. 12781–12789.doi: 10/f9d36r.
[175] Xiaojuan Yang, Xiaobin Zhao, Mitch A Phelps, Longzhu Piao, Darlene M Rozewski, Qing Liu,
L James Lee, Guido Marcucci, Michael R Grever, John C Byrd, et al. “A novel liposomal
formulation of flavopiridol”. In: International journal of pharmaceutics 365.1 (2009). Publisher:
Elsevier, pp. 170–174.doi: 10/bjbztv.
[176] Lu Zhang, Qinghua Chen, Yao Ma, and Jiashu Sun. “Microfluidic Methods for Fabrication and
Engineering of Nanoparticle Drug Delivery Systems”. In: ACS Applied Bio Materials 3.1 (2019).
Publisher: ACS Publications, pp. 107–120.doi: 10/gmpjg5.
[177] Qian-Xia Zhang, Li-Min Xu, Yue Zhou, Jie-Xin Wang, and Jian-Feng Chen. “Preparation of drug
nanoparticles using a T-junction microchannel system”. In: Industrial & engineering chemistry
research 50.24 (2011). Publisher: ACS Publications, pp. 13805–13812.doi: 10/b2t2rs.
101
[178] J. Zhao, R. Ong, G. Chen, Y. K. Juay, F. L. Ng, M. W. Lee, and C. H. Kua. “Development of rapid
manufacturing technology of polymer microfluidic devices by micro moulding using silicon
mould inserts”. In: Proceedings of the 6th International Conference on Nanochannels, Microchannels,
and Minichannels, ICNMM2008. Issue: PART B. American Society of Mechanical Engineers Digital
Collection, June 11, 2008, pp. 1187–1194.isbn: 0-7918-4834-5.doi: 10/fmzbb5.
[179] R. I. Zhdanov, O. V. Podobed, and V. V. Vlassov. “Cationic lipid-DNA complexes-lipoplexes-for
gene transfer and therapy”. In: Bioelectrochemistry (Amsterdam, Netherlands) 58.1 (Nov. 2002),
pp. 53–64.issn: 1567-5394.doi: 10/dbctr4.
[180] Igor V. Zhigaltsev, Nathan Belliveau, Ismail Hafez, Alex K. K. Leung, Jens Huft, Carl Hansen, and
Pieter R. Cullis. “Bottom-Up Design and Synthesis of Limit Size Lipid Nanoparticle Systems with
Aqueous and Triglyceride Cores Using Millisecond Microfluidic Mixing”. In: Langmuir 28.7
(Feb. 21, 2012), pp. 3633–3640.issn: 0743-7463, 1520-5827.doi: 10.1021/la204833h. (Visited on
07/11/2022).
[181] Weihong Zhong, Fan Li, Zuoguang Zhang, Lulu Song, and Zhimin Li. “Short fiber reinforced
composites for fused deposition modeling”. In: Materials Science and Engineering: A 301.2 (2001).
Publisher: Elsevier, pp. 125–130.doi: 10.1016/s0921-5093(00)01810-4.
[182] Jinwen Zhou, Amanda Vera Ellis, and Nicolas Hans Voelcker. “Recent developments in PDMS
surface modification for microfluidic devices”. In: Electrophoresis 31.1 (2010). Publisher: Wiley
Online Library, pp. 2–16.doi: 10/bk67pc.
[183] Feng Zhu, Timo Friedrich, Dayanthi Nugegoda, Jan Kaslin, and Donald Wlodkowic. “Assessment
of the biocompatibility of three-dimensional-printed polymers using multispecies toxicity tests”.
In: Biomicrofluidics 9.6 (2015), pp. 1–5.issn: 19321058.doi: 10.1063/1.4939031.
[184] Justin M Zook and Wyatt N Vreeland. “Effects of temperature, acyl chain length, and flow-rate
ratio on liposome formation and size in a microfluidic hydrodynamic focusing device”. In: Soft
Matter 6.6 (2010). Publisher: Royal Society of Chemistry, pp. 1352–1360.doi: 10.1039/b923299k.
102

Abstract (if available)

Abstract In recent years, microfluidic systems have found broad applications in biopharmaceutical processes. The unique advantages of microfluidics include precise fluid control, enhanced mixing, small footprint, and compatibility with fluid control equipment for automation. These qualities are particularly appealing to the biopharmaceutical community because of the inherently sophisticated biochemical procedures that requires precise control over precious analytical samples and synthetic precursors. For instance, well-controlled microscale mixing has attracted widespread interest for nanodrug fabrication. Despite having compelling potential for pharmaceutical applications, the industry has been slow in adopting microfluidics. This is largely due to the lengthy fabrication process and poor fluidic interfaces of traditional microfluidic devices. To bridge the gap between microfluidics research and industrial applications, the microfluidics community has taken to additive manufacturing (i.e. 3D printing) for microfluidic device fabrication because of its unprecedented ease of use, rapid prototyping, 3D capability for user-friendly fluidic interfaces, and low cost. Further exploration in utilizing 3D-printed microfluidic devices biopharmaceutical applications is critical to disseminate microfluidics beyond small-scale research settings.

This dissertation begins with a background on microfluidic technology, its applications in pharmaceutical development, and 3D printing for microfluidic device fabrication. The research work in the following chapters focus on microfluidic applications in nanodrug fabrication and affinity reagent generation. In chapter 2, microfluidic liposome production and concurrent loading of drug simulants is introduced. The precise control and enhanced mixing in microfluidics enable fine tuning of liposome sizes and drug encapsulation efficiencies by simply manipulating the flow parameters. In chapter 3, we introduce a novel, 3D-printed hydrodynamic flow focusing device that employs an elegant 4-way sheath flow channel and a downstream staggered herringbone mixer for producing mRNA-encapsulated lipid nanoparticles. At high throughput, the device produces size-limited lipid nanoparticles with high mRNA encapsulation efficiencies owing to its superior fluid focusing and rapid mixing capabilities. In chapter 4, a 3D-printed microfluidic device that enables kinetic off-rate selection for affinity reagent development is presented. By using a continuous flow technique, kinetic off-rate selection is achieved with improved selectivity for high affinity ligands. This work encourages the development of microfluidic platforms for automated affinity ligand screening. To explore the utility of 3D-printed microfluidics for biochemical reactions, we evaluated commercially-available, photocurable resins for stereolithography printing for their compatibility with fundamental in vitro enzymatic reactions, including PCR, transcription, translation, and reverse transcription. We found that most resins demonstrated low enzyme compatibility because of surface adsorption and toxic leachates that diffuse from the printed parts during the reaction. The closing chapter concludes this dissertation and provides future perspectives on 3D-printed microfluidics for the biopharmaceutical research community.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Expanded functionality and scalability of modular fluidic and instrumentation components

PDF

Synthesis of high-quality nanoparticles using microfluidic platforms

PDF

Rapid microfluidic prototyping for biological applications via stereolithographic 3-D printing: a journey

PDF

Applications of continuous flow reactors towards screening catalytically active nanoparticle syntheses

PDF

Cooperative effect of ionic liquid solvents in microfluidic devices for fabricating monodisperse metal nanoparticles

PDF

Scale-up of vapor-phase deposition of polymers: towards large-scale processing

PDF

Hybrid vat photopolymerization processes for viscous photocurable and non-photocurable materials

PDF

Statistical modeling and machine learning for shape accuracy control in additive manufacturing

PDF

Democratizing optical biopsy with diffuse optical imaging

PDF

Hybrid vat photopolymerization: methods and systems

PDF

Fabrication of ultrasound transducer and 3D-prinitng ultrasonic device

PDF

A modular microscale laboratory

PDF

Scalable polymerization additive manufacturing: principle and optimization

$Metasurfaces in 3D applications: multiscale stereolithography and inverse design of diffractive optical elements for structured light$

PDF

Metasurfaces in 3D applications: multiscale stereolithography and inverse design of diffractive optical elements for structured light

PDF

Fabrication-aware machine learning for accuracy control in additive manufacturing

PDF

Sustainable continuous flow syntheses of colloidal inorganic nanoparticle catalysts

PDF

2D ultrasonic transducer array’s design and fabrication with 3D printed interposer and applications

PDF

Selective Separation Shaping (SSS): large scale cementitious fabrication potentials

PDF

Machine learning-driven deformation prediction and compensation for additive manufacturing

PDF

Efficient delivery of augmented information services over distributed computing networks

Asset Metadata

Creator Lin, Wan-Zhen Sophie (author)

Core Title Intricate microfluidic devices for biopharmaceutical processes: forging ahead with additive manufacturing

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Chemical Engineering

Degree Conferral Date 2022-12

Publication Date 10/23/2023

Defense Date 09/12/2022

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag 3D printing,additive manufacturing,ligands,lipid nanoparticles,liposomes,microfluidics,mRNA display,nanodrug,OAI-PMH Harvest,selection,stereolithography

Format theses (aat)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Malmstadt, Noah (committee chair), Roberts, Richard W. (committee member), Zavaleta, Cristina (committee member)

Creator Email sophie@alumni.usc.edu,wanzhenl@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC112195797

Unique identifier UC112195797

Identifier etd-LinWanZhen-11281.pdf (filename)

Legacy Identifier etd-LinWanZhen-11281

Document Type Dissertation

Format theses (aat)

Rights Lin, Wan-Zhen Sophie

Internet Media Type application/pdf

Type texts

Source 20221026-usctheses-batch-988 (batch), University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu